Version not found.
\n \n Back to timeline\n \n\n This is the first version -- there is no previous version to compare against.\n
\n\n {prevVersion.id} ({prevVersion.loc} LOC) → {version} ({currentVersion.loc} LOC)\n
\n{prevMeta?.subtitle}
\n{prevVersion.loc} LOC
\n{prevVersion.tools.length} tools: {prevVersion.tools.join(\", \")}
\n{meta.subtitle}
\n{currentVersion.loc} LOC
\n{currentVersion.tools.length} tools: {currentVersion.tools.join(\", \")}
\n{version}
\n\n {meta.subtitle}\n
\n\n {meta.keyInsight}\n\n )}\n
{t(\"subtitle\")}
\n{metaA?.subtitle}
\n{infoA.loc} LOC
\n{infoA.tools.length} tools
\n {metaA &&{metaB?.subtitle}
\n{infoB.loc} LOC
\n{infoB.tools.length} tools
\n {metaB &&{t(\"none\")}
\n ) : (\n{t(\"none\")}
\n ) : (\n{t(\"none\")}
\n ) : (\n{t(\"empty_hint\")}
\n{t(\"subtitle\")}
\n\n {t(layer.id)}\n
\n\n {meta.subtitle}\n
\n )}\n\n {meta.keyInsight}\n
\n )}\n\n {t(\"subtitle\")}\n
\n\n {t(\"hero_subtitle\")}\n
\n\n {t(\"core_pattern_desc\")}\n
\n\n \n while\n \n True\n :\n {\"\\n\"}\n {\" \"}response = client.messages.\n create\n (\n messages=\n messages\n ,\n tools=\n tools\n )\n {\"\\n\"}\n {\" \"}if\n response.stop_reason != \n "tool_use"\n :\n {\"\\n\"}\n {\" \"}break\n {\"\\n\"}\n {\" \"}for\n tool_call \n in\n response.content\n :\n {\"\\n\"}\n {\" \"}result = \n execute_tool\n (\n tool_call.name\n ,\n tool_call.input\n )\n {\"\\n\"}\n {\" \"}messages.\n append\n (\n result\n )\n \n \n \n {t(\"message_flow_desc\")}\n
\n\n {t(\"learning_path_desc\")}\n
\n\n {meta.keyInsight}\n
\n \n \n );\n })}\n\n {t(\"layers_desc\")}\n
\n\n {description}\n
\n\n {decision.alternatives && (\n\n {decision.alternatives}\n
\n\n \n {lines.map((line, i) => (\n \n \n {i + 1}\n \n \n {highlightLine(line)}\n \n \n ))}\n \n \n | \n {row.oldNum ?? \"\"}\n | \n\n {row.newNum ?? \"\"}\n | \n\n {row.type === \"add\" && +}\n {row.type === \"remove\" && -}\n | \n\n | \n \n ))}\n
| \n {row.left.num ?? \"\"}\n | \n\n {row.left.text}\n | \n\n {row.right.num ?? \"\"}\n | \n\n {row.right.text}\n | \n
\n +{diff.locDelta} lines\n
\n/g,\n ''\n );\n\n // Wrap plain pre>code (ASCII art / diagrams) in diagram container\n html = html.replace(\n /]*)>/g,\n ''\n );\n\n // Mark the first blockquote as hero callout\n html = html.replace(\n //,\n ''\n );\n\n // Remove the h1 (it's redundant with the page header)\n html = html.replace(/.*?<\\/h1>\\n?/, \"\");\n\n // Fix ordered list counter for interrupted lists (ol start=\"N\")\n html = html.replace(\n //g,\n (_, start) => ``\n );\n\n return html;\n}\n\nexport function DocRenderer({ version }: DocRendererProps) {\n const locale = useLocale();\n\n const doc = useMemo(() => {\n const match = docsData.find(\n (d: { version: string; locale: string }) =>\n d.version === version && d.locale === locale\n );\n if (match) return match;\n return docsData.find(\n (d: { version: string; locale: string }) =>\n d.version === version && d.locale === \"en\"\n );\n }, [version, locale]);\n\n if (!doc) return null;\n\n const html = useMemo(() => {\n const raw = renderMarkdown(doc.content);\n return postProcessHtml(raw);\n }, [doc.content]);\n\n return (\n \n \n \n \n Learn Claude Code\n \n\n {/* Desktop nav */}\n \n\n {/* Mobile hamburger */}\n setMobileOpen(!mobileOpen)}\n className=\"flex min-h-[44px] min-w-[44px] items-center justify-center md:hidden\"\n >\n {mobileOpen ? \n\n {/* Mobile menu */}\n {mobileOpen && (\n \n {NAV_ITEMS.map((item) => (\n setMobileOpen(false)}\n >\n {t(item.key)}\n \n ))}\n \n \n {LOCALES.map((l) => (\n switchLocale(l.code)}\n className={cn(\n \"min-h-[44px] min-w-[44px] rounded-md px-3 text-xs font-medium\",\n locale === l.code\n ? \"bg-zinc-900 text-white dark:bg-white dark:text-zinc-900\"\n : \"border border-[var(--color-border)]\"\n )}\n >\n {l.label}\n \n ))}\n \n \n \n {dark ? \n \n \n \n )}\n \n );\n}\n"
},
{
"path": "web/src/components/layout/sidebar.tsx",
"content": "\"use client\";\n\nimport Link from \"next/link\";\nimport { usePathname } from \"next/navigation\";\nimport { LAYERS, VERSION_META } from \"@/lib/constants\";\nimport { useTranslations } from \"@/lib/i18n\";\nimport { cn } from \"@/lib/utils\";\n\nconst LAYER_DOT_BG: Record = {\n tools: \"bg-blue-500\",\n planning: \"bg-emerald-500\",\n memory: \"bg-purple-500\",\n concurrency: \"bg-amber-500\",\n collaboration: \"bg-red-500\",\n};\n\nexport function Sidebar() {\n const pathname = usePathname();\n const locale = pathname.split(\"/\")[1] || \"en\";\n const t = useTranslations(\"sessions\");\n const tLayer = useTranslations(\"layer_labels\");\n\n return (\n \n );\n}\n"
},
{
"path": "web/src/components/simulator/agent-loop-simulator.tsx",
"content": "\"use client\";\n\nimport { useRef, useEffect, useState } from \"react\";\nimport { AnimatePresence } from \"framer-motion\";\nimport { useTranslations } from \"@/lib/i18n\";\nimport { useSimulator } from \"@/hooks/useSimulator\";\nimport { SimulatorControls } from \"./simulator-controls\";\nimport { SimulatorMessage } from \"./simulator-message\";\nimport type { Scenario } from \"@/types/agent-data\";\n\nconst scenarioModules: Record Promise<{ default: Scenario }>> = {\n s01: () => import(\"@/data/scenarios/s01.json\") as Promise<{ default: Scenario }>,\n s02: () => import(\"@/data/scenarios/s02.json\") as Promise<{ default: Scenario }>,\n s03: () => import(\"@/data/scenarios/s03.json\") as Promise<{ default: Scenario }>,\n s04: () => import(\"@/data/scenarios/s04.json\") as Promise<{ default: Scenario }>,\n s05: () => import(\"@/data/scenarios/s05.json\") as Promise<{ default: Scenario }>,\n s06: () => import(\"@/data/scenarios/s06.json\") as Promise<{ default: Scenario }>,\n s07: () => import(\"@/data/scenarios/s07.json\") as Promise<{ default: Scenario }>,\n s08: () => import(\"@/data/scenarios/s08.json\") as Promise<{ default: Scenario }>,\n s09: () => import(\"@/data/scenarios/s09.json\") as Promise<{ default: Scenario }>,\n s10: () => import(\"@/data/scenarios/s10.json\") as Promise<{ default: Scenario }>,\n s11: () => import(\"@/data/scenarios/s11.json\") as Promise<{ default: Scenario }>,\n s12: () => import(\"@/data/scenarios/s12.json\") as Promise<{ default: Scenario }>,\n};\n\ninterface AgentLoopSimulatorProps {\n version: string;\n}\n\nexport function AgentLoopSimulator({ version }: AgentLoopSimulatorProps) {\n const t = useTranslations(\"version\");\n const [scenario, setScenario] = useState(null);\n const scrollRef = useRef(null);\n\n useEffect(() => {\n const loader = scenarioModules[version];\n if (loader) {\n loader().then((mod) => setScenario(mod.default));\n }\n }, [version]);\n\n const sim = useSimulator(scenario?.steps ?? []);\n\n useEffect(() => {\n if (scrollRef.current) {\n scrollRef.current.scrollTo({\n top: scrollRef.current.scrollHeight,\n behavior: \"smooth\",\n });\n }\n }, [sim.visibleSteps.length]);\n\n if (!scenario) return null;\n\n return (\n \n {t(\"simulator\")}
\n \n {scenario.description}\n
\n\n \n \n \n {sim.visibleSteps.length === 0 && (\n \n Press Play or Step to begin\n \n )}\n \n {sim.visibleSteps.map((step, i) => (\n \n \n \n \n );\n}\n"
},
{
"path": "web/src/components/simulator/simulator-controls.tsx",
"content": "\"use client\";\n\nimport { useTranslations } from \"@/lib/i18n\";\nimport { Play, Pause, SkipForward, RotateCcw } from \"lucide-react\";\nimport { cn } from \"@/lib/utils\";\n\ninterface SimulatorControlsProps {\n isPlaying: boolean;\n isComplete: boolean;\n currentIndex: number;\n totalSteps: number;\n speed: number;\n onPlay: () => void;\n onPause: () => void;\n onStep: () => void;\n onReset: () => void;\n onSpeedChange: (speed: number) => void;\n}\n\nconst SPEEDS = [0.5, 1, 2, 4];\n\nexport function SimulatorControls({\n isPlaying,\n isComplete,\n currentIndex,\n totalSteps,\n speed,\n onPlay,\n onPause,\n onStep,\n onReset,\n onSpeedChange,\n}: SimulatorControlsProps) {\n const t = useTranslations(\"sim\");\n\n return (\n \n \n {isPlaying ? (\n \n \n \n \n \n\n \n \n {t(\"speed\")}:\n \n {SPEEDS.map((s) => (\n onSpeedChange(s)}\n className={cn(\n \"rounded px-2 py-1 text-xs font-medium transition-colors\",\n speed === s\n ? \"bg-zinc-900 text-white dark:bg-white dark:text-zinc-900\"\n : \"text-[var(--color-text-secondary)] hover:text-[var(--color-text)]\"\n )}\n >\n {s}x\n \n ))}\n \n\n \n {Math.max(0, currentIndex + 1)} {t(\"step_of\")} {totalSteps}\n \n \n );\n}\n"
},
{
"path": "web/src/components/simulator/simulator-message.tsx",
"content": "\"use client\";\n\nimport { motion } from \"framer-motion\";\nimport { cn } from \"@/lib/utils\";\nimport type { SimStep } from \"@/types/agent-data\";\nimport { User, Bot, Terminal, ArrowRight, AlertCircle } from \"lucide-react\";\n\ninterface SimulatorMessageProps {\n step: SimStep;\n index: number;\n}\n\nconst TYPE_CONFIG: Record<\n string,\n { icon: typeof User; label: string; bgClass: string; borderClass: string }\n> = {\n user_message: {\n icon: User,\n label: \"User\",\n bgClass: \"bg-blue-50 dark:bg-blue-950/30\",\n borderClass: \"border-blue-200 dark:border-blue-800\",\n },\n assistant_text: {\n icon: Bot,\n label: \"Assistant\",\n bgClass: \"bg-zinc-50 dark:bg-zinc-900\",\n borderClass: \"border-zinc-200 dark:border-zinc-700\",\n },\n tool_call: {\n icon: Terminal,\n label: \"Tool Call\",\n bgClass: \"bg-amber-50 dark:bg-amber-950/30\",\n borderClass: \"border-amber-200 dark:border-amber-800\",\n },\n tool_result: {\n icon: ArrowRight,\n label: \"Tool Result\",\n bgClass: \"bg-emerald-50 dark:bg-emerald-950/30\",\n borderClass: \"border-emerald-200 dark:border-emerald-800\",\n },\n system_event: {\n icon: AlertCircle,\n label: \"System\",\n bgClass: \"bg-purple-50 dark:bg-purple-950/30\",\n borderClass: \"border-purple-200 dark:border-purple-800\",\n },\n};\n\nexport function SimulatorMessage({ step, index }: SimulatorMessageProps) {\n const config = TYPE_CONFIG[step.type] || TYPE_CONFIG.assistant_text;\n const Icon = config.icon;\n\n return (\n \n \n \n {step.content || \"(empty)\"}\n \n ) : step.type === \"system_event\" ? (\n \n {step.content}\n \n ) : (\n {step.content}
\n )}\n\n \n {step.annotation}\n
\n \n );\n}\n"
},
{
"path": "web/src/components/timeline/timeline.tsx",
"content": "\"use client\";\n\nimport Link from \"next/link\";\nimport { motion } from \"framer-motion\";\nimport { useTranslations, useLocale } from \"@/lib/i18n\";\nimport { LEARNING_PATH, VERSION_META, LAYERS } from \"@/lib/constants\";\nimport { LayerBadge } from \"@/components/ui/badge\";\nimport { cn } from \"@/lib/utils\";\nimport versionsData from \"@/data/generated/versions.json\";\n\nconst LAYER_DOT_BG: Record = {\n tools: \"bg-blue-500\",\n planning: \"bg-emerald-500\",\n memory: \"bg-purple-500\",\n concurrency: \"bg-amber-500\",\n collaboration: \"bg-red-500\",\n};\n\nconst LAYER_LINE_BG: Record = {\n tools: \"bg-blue-500/30\",\n planning: \"bg-emerald-500/30\",\n memory: \"bg-purple-500/30\",\n concurrency: \"bg-amber-500/30\",\n collaboration: \"bg-red-500/30\",\n};\n\nconst LAYER_BAR_BG: Record = {\n tools: \"bg-blue-500\",\n planning: \"bg-emerald-500\",\n memory: \"bg-purple-500\",\n concurrency: \"bg-amber-500\",\n collaboration: \"bg-red-500\",\n};\n\nfunction getVersionData(id: string) {\n return versionsData.versions.find((v) => v.id === id);\n}\n\nconst MAX_LOC = Math.max(\n ...versionsData.versions\n .filter((v) => LEARNING_PATH.includes(v.id as (typeof LEARNING_PATH)[number]))\n .map((v) => v.loc)\n);\n\nexport function Timeline() {\n const t = useTranslations(\"timeline\");\n const tv = useTranslations(\"version\");\n const locale = useLocale();\n\n return (\n \n {/* Layer Legend */}\n \n \n {t(\"layer_legend\")}\n
\n \n {LAYERS.map((layer) => (\n \n \n \n {versionId.replace(\"s\", \"\").replace(\"_mini\", \"m\")}\n \n \n {!isLast && (\n \n \n {versionId} \n \n {meta.coreAddition}\n \n \n\n \n {meta.title}\n \n {meta.subtitle}\n \n
\n\n {/* Stats row */}\n \n \n {data.loc} {tv(\"loc\")}\n \n \n {data.tools.length} {tv(\"tools\")}\n \n \n\n {/* LOC bar */}\n \n \n “{meta.keyInsight}”\n
\n )}\n\n {/* Link */}\n \n {t(\"learn_more\")}\n →\n \n \n \n \n );\n })}\n \n\n {/* LOC Growth Chart */}\n \n {t(\"loc_growth\")}
\n \n {LEARNING_PATH.map((versionId) => {\n const meta = VERSION_META[versionId];\n const data = getVersionData(versionId);\n if (!meta || !data) return null;\n\n const widthPercent = Math.max(\n 2,\n Math.round((data.loc / MAX_LOC) * 100)\n );\n\n return (\n \n \n {versionId}\n \n \n \n \n \n {data.loc}\n \n \n \n \n \n );\n })}\n \n \n \n );\n}\n"
},
{
"path": "web/src/components/ui/badge.tsx",
"content": "import { cn } from \"@/lib/utils\";\n\nconst LAYER_COLORS = {\n tools:\n \"bg-blue-100 text-blue-800 dark:bg-blue-900/30 dark:text-blue-300\",\n planning:\n \"bg-emerald-100 text-emerald-800 dark:bg-emerald-900/30 dark:text-emerald-300\",\n memory:\n \"bg-purple-100 text-purple-800 dark:bg-purple-900/30 dark:text-purple-300\",\n concurrency:\n \"bg-amber-100 text-amber-800 dark:bg-amber-900/30 dark:text-amber-300\",\n collaboration:\n \"bg-red-100 text-red-800 dark:bg-red-900/30 dark:text-red-300\",\n} as const;\n\ninterface BadgeProps {\n layer: keyof typeof LAYER_COLORS;\n children: React.ReactNode;\n className?: string;\n}\n\nexport function LayerBadge({ layer, children, className }: BadgeProps) {\n return (\n \n {children}\n \n );\n}\n"
},
{
"path": "web/src/components/ui/card.tsx",
"content": "import { cn } from \"@/lib/utils\";\n\ninterface CardProps extends React.HTMLAttributes {\n children: React.ReactNode;\n}\n\nexport function Card({ className, children, ...props }: CardProps) {\n return (\n \n {children}\n \n );\n}\n\nexport function CardHeader({ className, children, ...props }: CardProps) {\n return (\n \n {children}\n \n );\n}\n\nexport function CardTitle({\n className,\n children,\n ...props\n}: React.HTMLAttributes) {\n return (\n \n {children}\n
\n );\n}\n"
},
{
"path": "web/src/components/ui/tabs.tsx",
"content": "\"use client\";\n\nimport { useState } from \"react\";\nimport { cn } from \"@/lib/utils\";\n\ninterface TabsProps {\n tabs: { id: string; label: string }[];\n defaultTab?: string;\n children: (activeTab: string) => React.ReactNode;\n className?: string;\n}\n\nexport function Tabs({ tabs, defaultTab, children, className }: TabsProps) {\n const [active, setActive] = useState(defaultTab || tabs[0]?.id || \"\");\n\n return (\n \n \n {tabs.map((tab) => (\n setActive(tab.id)}\n className={cn(\n \"px-4 py-2 text-sm font-medium transition-colors\",\n active === tab.id\n ? \"border-b-2 border-zinc-900 text-zinc-900 dark:border-white dark:text-white\"\n : \"text-zinc-500 hover:text-zinc-700 dark:text-zinc-400 dark:hover:text-zinc-200\"\n )}\n >\n {tab.label}\n \n ))}\n \n {children(active)}\n \n );\n}\n"
},
{
"path": "web/src/components/visualizations/index.tsx",
"content": "\"use client\";\n\nimport { lazy, Suspense } from \"react\";\nimport { useTranslations } from \"@/lib/i18n\";\n\nconst visualizations: Record<\n string,\n React.LazyExoticComponent>\n> = {\n s01: lazy(() => import(\"./s01-agent-loop\")),\n s02: lazy(() => import(\"./s02-tool-dispatch\")),\n s03: lazy(() => import(\"./s03-todo-write\")),\n s04: lazy(() => import(\"./s04-subagent\")),\n s05: lazy(() => import(\"./s05-skill-loading\")),\n s06: lazy(() => import(\"./s06-context-compact\")),\n s07: lazy(() => import(\"./s07-task-system\")),\n s08: lazy(() => import(\"./s08-background-tasks\")),\n s09: lazy(() => import(\"./s09-agent-teams\")),\n s10: lazy(() => import(\"./s10-team-protocols\")),\n s11: lazy(() => import(\"./s11-autonomous-agents\")),\n s12: lazy(() => import(\"./s12-worktree-task-isolation\")),\n};\n\nexport function SessionVisualization({ version }: { version: string }) {\n const t = useTranslations(\"viz\");\n const Component = visualizations[version];\n if (!Component) return null;\n return (\n \n \n {title || \"The Agent While-Loop\"}\n
\n\n \n \n {/* Left panel: SVG Flowchart (60%) */}\n \n \n while (stop_reason === \"tool_use\")\n \n \n \n \n \n \n \n \n \n \n\n {/* Edges */}\n {EDGES.map((edge) => {\n const key = `${edge.from}->${edge.to}`;\n const isActive = activeEdges.includes(key);\n const d = edgePath(edge.from, edge.to);\n\n return (\n \n \n {edge.label}\n \n )}\n \n );\n })}\n\n {/* Nodes */}\n {NODES.map((node) => {\n const isActive = activeNodes.includes(node.id);\n const isEnd = node.id === \"end\";\n const filterAttr = isActive\n ? isEnd\n ? \"url(#glow-purple)\"\n : \"url(#glow-blue)\"\n : \"none\";\n\n if (node.type === \"diamond\") {\n // Diamond shape for decision node\n const cx = node.x;\n const cy = node.y;\n const hw = node.w / 2;\n const hh = node.h / 2;\n const points = `${cx},${cy - hh} ${cx + hw},${cy} ${cx},${cy + hh} ${cx - hw},${cy}`;\n return (\n \n \n {node.label}\n \n \n );\n }\n\n return (\n \n \n {node.label}\n \n \n );\n })}\n\n {/* Iteration counter */}\n {currentStep >= 5 && (\n \n iter #2\n \n )}\n \n \n\n {/* Right panel: messages[] array (40%) */}\n \n \n messages[]\n \n \n \n {visibleMessages.length === 0 && (\n \n [ empty ]\n \n )}\n {visibleMessages.map((msg, i) => (\n \n \n {msg.role}\n \n \n {msg.detail}\n \n \n ))}\n \n\n {/* Array index markers */}\n {visibleMessages.length > 0 && (\n \n \n length: {visibleMessages.length}\n \n \n )}\n \n \n \n \n\n \n );\n}\n"
},
{
"path": "web/src/components/visualizations/s02-tool-dispatch.tsx",
"content": "\"use client\";\n\nimport { motion, AnimatePresence } from \"framer-motion\";\nimport { useSteppedVisualization } from \"@/hooks/useSteppedVisualization\";\nimport { StepControls } from \"@/components/visualizations/shared/step-controls\";\nimport { useSvgPalette } from \"@/hooks/useDarkMode\";\n\n// -- Tool definitions --\n\ninterface ToolDef {\n name: string;\n desc: string;\n color: string;\n activeColor: string;\n darkColor: string;\n darkActiveColor: string;\n}\n\nconst TOOLS: ToolDef[] = [\n {\n name: \"bash\",\n desc: \"Execute shell commands\",\n color: \"border-orange-300 bg-orange-50\",\n activeColor: \"border-orange-500 bg-orange-100 ring-2 ring-orange-400\",\n darkColor: \"dark:border-zinc-700 dark:bg-zinc-800/50\",\n darkActiveColor: \"dark:border-orange-500 dark:bg-orange-950/40 dark:ring-orange-500\",\n },\n {\n name: \"read_file\",\n desc: \"Read file contents\",\n color: \"border-sky-300 bg-sky-50\",\n activeColor: \"border-sky-500 bg-sky-100 ring-2 ring-sky-400\",\n darkColor: \"dark:border-zinc-700 dark:bg-zinc-800/50\",\n darkActiveColor: \"dark:border-sky-500 dark:bg-sky-950/40 dark:ring-sky-500\",\n },\n {\n name: \"write_file\",\n desc: \"Create or overwrite a file\",\n color: \"border-emerald-300 bg-emerald-50\",\n activeColor: \"border-emerald-500 bg-emerald-100 ring-2 ring-emerald-400\",\n darkColor: \"dark:border-zinc-700 dark:bg-zinc-800/50\",\n darkActiveColor: \"dark:border-emerald-500 dark:bg-emerald-950/40 dark:ring-emerald-500\",\n },\n {\n name: \"edit_file\",\n desc: \"Apply targeted edits\",\n color: \"border-violet-300 bg-violet-50\",\n activeColor: \"border-violet-500 bg-violet-100 ring-2 ring-violet-400\",\n darkColor: \"dark:border-zinc-700 dark:bg-zinc-800/50\",\n darkActiveColor: \"dark:border-violet-500 dark:bg-violet-950/40 dark:ring-violet-500\",\n },\n];\n\n// Per-step: which tool index is active (-1 = none, 4 = all)\nconst ACTIVE_TOOL_PER_STEP: number[] = [-1, 0, 1, 2, 3, 4];\n\n// Incoming request JSON per step\nconst REQUEST_PER_STEP: (string | null)[] = [\n null,\n '{ name: \"bash\", input: { cmd: \"ls -la\" } }',\n '{ name: \"read_file\", input: { path: \"src/auth.ts\" } }',\n '{ name: \"write_file\", input: { path: \"config.json\" } }',\n '{ name: \"edit_file\", input: { path: \"index.ts\" } }',\n null,\n];\n\n// Step annotations\nconst STEP_INFO = [\n { title: \"The Dispatch Map\", desc: \"A dictionary maps tool names to handler functions. The loop code never changes.\" },\n { title: \"Route: bash\", desc: \"tool_call.name -> handlers['bash'](input). Name-based routing.\" },\n { title: \"Route: read_file\", desc: \"Same pattern, different handler. Validate input, execute, return result.\" },\n { title: \"Route: write_file\", desc: \"Every tool returns a tool_result that goes back into messages[].\" },\n { title: \"Route: edit_file\", desc: \"Adding a new tool = adding one entry to the dispatch map.\" },\n { title: \"The Key Insight\", desc: \"The while loop stays the same. You only grow the dispatch map. That's it.\" },\n];\n\n// SVG layout constants\nconst SVG_WIDTH = 600;\nconst SVG_HEIGHT = 320;\nconst DISPATCHER_X = SVG_WIDTH / 2;\nconst DISPATCHER_Y = 60;\nconst DISPATCHER_W = 160;\nconst DISPATCHER_H = 50;\nconst CARD_Y = 230;\nconst CARD_W = 110;\nconst CARD_H = 65;\nconst CARD_GAP = 20;\n\nfunction getCardX(index: number): number {\n const totalWidth = TOOLS.length * CARD_W + (TOOLS.length - 1) * CARD_GAP;\n const startX = (SVG_WIDTH - totalWidth) / 2;\n return startX + index * (CARD_W + CARD_GAP) + CARD_W / 2;\n}\n\nexport default function ToolDispatch({ title }: { title?: string }) {\n const {\n currentStep,\n totalSteps,\n next,\n prev,\n reset,\n isPlaying,\n toggleAutoPlay,\n } = useSteppedVisualization({ totalSteps: 6, autoPlayInterval: 2500 });\n\n const palette = useSvgPalette();\n const activeToolIdx = ACTIVE_TOOL_PER_STEP[currentStep];\n const request = REQUEST_PER_STEP[currentStep];\n const stepInfo = STEP_INFO[currentStep];\n const isAllActive = activeToolIdx === 4;\n\n return (\n \n \n {title || \"Tool Dispatch Map\"}\n
\n\n \n {/* Incoming request display */}\n \n \n Incoming:\n \n \n {request && (\n \n {request}\n \n )}\n {!request && currentStep === 0 && (\n \n waiting for tool_call...\n \n )}\n {isAllActive && (\n \n All routes active\n \n )}\n \n \n\n {/* SVG dispatch diagram */}\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n\n {/* Dispatcher box */}\n 0 ? palette.activeNodeFill : palette.nodeFill,\n stroke: currentStep > 0 ? palette.activeNodeStroke : palette.nodeStroke,\n }}\n filter={currentStep > 0 ? \"url(#dispatch-glow)\" : \"none\"}\n transition={{ duration: 0.4 }}\n />\n 0 ? palette.activeNodeText : palette.nodeText }}\n transition={{ duration: 0.4 }}\n >\n dispatch(name)\n \n\n {/* Connection lines from dispatcher to each tool card */}\n {TOOLS.map((tool, i) => {\n const cardX = getCardX(i);\n const isActive = isAllActive || i === activeToolIdx;\n const lineColor = isActive ? palette.activeEdgeStroke : palette.edgeStroke;\n\n return (\n \n \n {tool.name}\n \n \n {tool.desc}\n \n \n );\n })}\n\n {/* \"+\" extensibility indicator on last step */}\n {isAllActive && (\n \n \n +\n \n \n )}\n \n\n {/* Code snippet below the diagram */}\n \n \n const handlers = {\"{\"}\n {TOOLS.map((tool, i) => {\n const isActive = isAllActive || i === activeToolIdx;\n return (\n \n {\" \"}{tool.name},\n \n );\n })}\n {\" }{\"}{\"}\"};\n \n \n \n\n \n );\n}\n"
},
{
"path": "web/src/components/visualizations/s03-todo-write.tsx",
"content": "\"use client\";\n\nimport { motion, AnimatePresence } from \"framer-motion\";\nimport { useSteppedVisualization } from \"@/hooks/useSteppedVisualization\";\nimport { StepControls } from \"@/components/visualizations/shared/step-controls\";\n\n// -- Task definitions --\n\ntype TaskStatus = \"pending\" | \"in_progress\" | \"done\";\n\ninterface Task {\n id: number;\n label: string;\n status: TaskStatus;\n}\n\n// Snapshot of all 4 tasks at each step\nconst TASK_STATES: Task[][] = [\n // Step 0: all pending\n [\n { id: 1, label: \"Write auth tests\", status: \"pending\" },\n { id: 2, label: \"Fix mobile layout\", status: \"pending\" },\n { id: 3, label: \"Add error handling\", status: \"pending\" },\n { id: 4, label: \"Update config loader\", status: \"pending\" },\n ],\n // Step 1: still all pending (idle round 1)\n [\n { id: 1, label: \"Write auth tests\", status: \"pending\" },\n { id: 2, label: \"Fix mobile layout\", status: \"pending\" },\n { id: 3, label: \"Add error handling\", status: \"pending\" },\n { id: 4, label: \"Update config loader\", status: \"pending\" },\n ],\n // Step 2: still all pending (idle round 2)\n [\n { id: 1, label: \"Write auth tests\", status: \"pending\" },\n { id: 2, label: \"Fix mobile layout\", status: \"pending\" },\n { id: 3, label: \"Add error handling\", status: \"pending\" },\n { id: 4, label: \"Update config loader\", status: \"pending\" },\n ],\n // Step 3: NAG fires, task 1 moves to in_progress\n [\n { id: 1, label: \"Write auth tests\", status: \"in_progress\" },\n { id: 2, label: \"Fix mobile layout\", status: \"pending\" },\n { id: 3, label: \"Add error handling\", status: \"pending\" },\n { id: 4, label: \"Update config loader\", status: \"pending\" },\n ],\n // Step 4: task 1 done\n [\n { id: 1, label: \"Write auth tests\", status: \"done\" },\n { id: 2, label: \"Fix mobile layout\", status: \"pending\" },\n { id: 3, label: \"Add error handling\", status: \"pending\" },\n { id: 4, label: \"Update config loader\", status: \"pending\" },\n ],\n // Step 5: task 2 self-directed to in_progress\n [\n { id: 1, label: \"Write auth tests\", status: \"done\" },\n { id: 2, label: \"Fix mobile layout\", status: \"in_progress\" },\n { id: 3, label: \"Add error handling\", status: \"pending\" },\n { id: 4, label: \"Update config loader\", status: \"pending\" },\n ],\n // Step 6: tasks 2,3 done, task 4 in_progress\n [\n { id: 1, label: \"Write auth tests\", status: \"done\" },\n { id: 2, label: \"Fix mobile layout\", status: \"done\" },\n { id: 3, label: \"Add error handling\", status: \"done\" },\n { id: 4, label: \"Update config loader\", status: \"in_progress\" },\n ],\n];\n\n// Nag timer value at each step (out of 3)\nconst NAG_TIMER_PER_STEP = [0, 1, 2, 3, 0, 0, 0];\nconst NAG_THRESHOLD = 3;\n\n// Whether the nag fires at this step\nconst NAG_FIRES_PER_STEP = [false, false, false, true, false, false, false];\n\n// Step annotations\nconst STEP_INFO = [\n { title: \"The Plan\", desc: \"TodoWrite gives the model a visible plan. All tasks start as pending.\" },\n { title: \"Round 1 -- Idle\", desc: \"The model does work but doesn't touch its todos. The nag counter increments.\" },\n { title: \"Round 2 -- Still Idle\", desc: \"Two rounds without progress. Pressure builds.\" },\n { title: \"NAG!\", desc: \"Threshold reached! System message injected: 'You have pending tasks. Pick one up now!'\" },\n { title: \"Task Complete\", desc: \"The model completes the task. Timer stays at 0 -- working on todos resets the counter.\" },\n { title: \"Self-Directed\", desc: \"Once the model learns the pattern, it picks up tasks voluntarily.\" },\n { title: \"Mission Accomplished\", desc: \"Visible plan + nag pressure = reliable task completion.\" },\n];\n\n// -- Column component --\n\nfunction KanbanColumn({\n title,\n tasks,\n accentClass,\n headerBg,\n}: {\n title: string;\n tasks: Task[];\n accentClass: string;\n headerBg: string;\n}) {\n return (\n \n \n {title}\n \n {tasks.length}\n \n \n \n \n {tasks.map((task) => (\n \n {tasks.length === 0 && (\n \n --\n \n )}\n \n \n );\n}\n\n// -- Task card --\n\nfunction TaskCard({ task }: { task: Task }) {\n const statusStyles: Record = {\n pending: \"bg-zinc-100 text-zinc-600 dark:bg-zinc-800 dark:text-zinc-400\",\n in_progress: \"bg-amber-100 text-amber-700 dark:bg-amber-900/40 dark:text-amber-300\",\n done: \"bg-emerald-100 text-emerald-700 dark:bg-emerald-900/40 dark:text-emerald-300\",\n };\n\n const borderStyles: Record = {\n pending: \"border-zinc-200 bg-white dark:border-zinc-700 dark:bg-zinc-800\",\n in_progress: \"border-amber-300 bg-amber-50 dark:border-amber-700 dark:bg-amber-950/30\",\n done: \"border-emerald-300 bg-emerald-50 dark:border-emerald-700 dark:bg-emerald-950/30\",\n };\n\n return (\n \n \n \n #{task.id}\n \n \n {task.status.replace(\"_\", \" \")}\n \n \n \n {task.label}\n \n \n );\n}\n\n// -- Nag gauge --\n\nfunction NagGauge({ value, max, firing }: { value: number; max: number; firing: boolean }) {\n const pct = Math.min((value / max) * 100, 100);\n\n const barColor =\n value === 0\n ? \"bg-zinc-300 dark:bg-zinc-600\"\n : value === 1\n ? \"bg-green-400 dark:bg-green-500\"\n : value === 2\n ? \"bg-yellow-400 dark:bg-yellow-500\"\n : \"bg-red-500 dark:bg-red-500\";\n\n return (\n \n \n \n Nag Timer\n \n \n {value}/{max}\n \n \n \n \n \n {title || \"TodoWrite Nag System\"}\n
\n\n \n {/* Nag gauge + nag message */}\n \n \n {nagFires && (\n \n SYSTEM: \"You have pending tasks. Pick one up now!\"\n \n )}\n \n \n\n {/* Kanban board */}\n \n \n );\n}\n"
},
{
"path": "web/src/components/visualizations/s04-subagent.tsx",
"content": "\"use client\";\n\nimport { motion, AnimatePresence } from \"framer-motion\";\nimport { useSteppedVisualization } from \"@/hooks/useSteppedVisualization\";\nimport { StepControls } from \"@/components/visualizations/shared/step-controls\";\n\ninterface MessageBlock {\n id: string;\n label: string;\n color: string;\n}\n\nconst PARENT_BASE_MESSAGES: MessageBlock[] = [\n { id: \"p1\", label: \"user: Build login + tests\", color: \"bg-blue-500\" },\n { id: \"p2\", label: \"assistant: Planning approach...\", color: \"bg-zinc-600\" },\n { id: \"p3\", label: \"tool_result: project structure\", color: \"bg-emerald-500\" },\n];\n\nconst TASK_PROMPT: MessageBlock = {\n id: \"task\",\n label: \"task: Write unit tests for auth\",\n color: \"bg-purple-500\",\n};\n\nconst CHILD_WORK_MESSAGES: MessageBlock[] = [\n { id: \"c1\", label: \"tool_use: read auth.ts\", color: \"bg-amber-500\" },\n { id: \"c2\", label: \"tool_use: write test.ts\", color: \"bg-amber-500\" },\n];\n\nconst SUMMARY_BLOCK: MessageBlock = {\n id: \"summary\",\n label: \"summary: 3 tests written, all passing\",\n color: \"bg-teal-500\",\n};\n\nconst STEPS = [\n {\n title: \"Parent Context\",\n description:\n \"The parent agent has accumulated messages from the conversation.\",\n },\n {\n title: \"Spawn Subagent\",\n description:\n \"Task tool creates a child with fresh messages[]. Only the task description is passed.\",\n },\n {\n title: \"Independent Work\",\n description:\n \"The child has its own context. It doesn't see the parent's history.\",\n },\n {\n title: \"Compress Result\",\n description:\n \"The child's full conversation compresses into one summary.\",\n },\n {\n title: \"Return Summary\",\n description:\n \"Only the summary returns. The child's full context is discarded.\",\n },\n {\n title: \"Clean Context\",\n description:\n \"The parent gets a clean summary without context bloat. This is fresh-context isolation via messages[].\",\n },\n];\n\nexport default function SubagentIsolation({ title }: { title?: string }) {\n const {\n currentStep,\n totalSteps,\n next,\n prev,\n reset,\n isPlaying,\n toggleAutoPlay,\n } = useSteppedVisualization({ totalSteps: STEPS.length, autoPlayInterval: 2500 });\n\n // Derive what to show in each container based on step\n const parentMessages: MessageBlock[] = (() => {\n const base = [...PARENT_BASE_MESSAGES];\n if (currentStep >= 5) {\n base.push(SUMMARY_BLOCK);\n }\n return base;\n })();\n\n const childMessages: MessageBlock[] = (() => {\n if (currentStep < 1) return [];\n if (currentStep === 1) return [TASK_PROMPT];\n if (currentStep === 2) return [TASK_PROMPT, ...CHILD_WORK_MESSAGES];\n if (currentStep === 3) return [SUMMARY_BLOCK];\n return currentStep >= 4 ? [TASK_PROMPT, ...CHILD_WORK_MESSAGES] : [];\n })();\n\n const showChildEmpty = currentStep === 0;\n const showArcToChild = currentStep === 1;\n const showCompression = currentStep === 3;\n const showArcToParent = currentStep === 4;\n const childDiscarded = currentStep >= 4;\n const childFaded = currentStep >= 4;\n\n return (\n \n \n {title || \"Subagent Context Isolation\"}\n
\n\n \n {/* Main layout: two containers side by side */}\n \n {/* Parent Process Container */}\n \n \n \n \n Parent Process\n \n \n \n messages[]\n \n \n \n {parentMessages.map((msg, i) => (\n \n {msg.label}\n \n ))}\n \n \n {currentStep >= 5 && (\n \n 3 original + 1 summary = clean context\n \n )}\n \n\n {/* Isolation Wall */}\n \n \n = 1 && currentStep <= 4 ? 1 : 0.4,\n }}\n className=\"rounded bg-zinc-200 px-2 py-1 text-center font-mono text-[10px] text-zinc-500 dark:bg-zinc-700 dark:text-zinc-400\"\n style={{ writingMode: \"vertical-rl\", textOrientation: \"mixed\" }}\n >\n ISOLATION\n \n \n \n\n {/* Child Process Container */}\n \n \n \n Child Process\n \n \n \n messages[] (fresh)\n \n\n {showChildEmpty && (\n \n \n not yet spawned\n \n \n )}\n\n \n \n {childMessages.map((msg) => (\n \n {msg.label}\n \n ))}\n \n \n\n {showCompression && (\n \n Compressing full context into summary...\n \n )}\n\n {childDiscarded && (\n \n context discarded\n \n )}\n \n\n {/* Animated arcs: task prompt going from parent to child */}\n \n {showArcToChild && (\n \n \n task prompt\n \n \n )}\n \n\n \n {showArcToParent && (\n \n \n summary\n \n \n )}\n \n \n\n {/* Step Controls */}\n \n \n );\n}\n"
},
{
"path": "web/src/components/visualizations/s05-skill-loading.tsx",
"content": "\"use client\";\n\nimport { motion, AnimatePresence } from \"framer-motion\";\nimport { useSteppedVisualization } from \"@/hooks/useSteppedVisualization\";\nimport { StepControls } from \"@/components/visualizations/shared/step-controls\";\n\ninterface SkillEntry {\n name: string;\n summary: string;\n fullTokens: number;\n content: string[];\n}\n\nconst SKILLS: SkillEntry[] = [\n {\n name: \"/commit\",\n summary: \"Create git commits following repo conventions\",\n fullTokens: 320,\n content: [\n \"1. Run git status + git diff to see changes\",\n \"2. Analyze all staged changes and draft message\",\n \"3. Create commit with Co-Authored-By trailer\",\n \"4. Run git status after commit to verify\",\n ],\n },\n {\n name: \"/review-pr\",\n summary: \"Review pull requests for bugs and style\",\n fullTokens: 480,\n content: [\n \"1. Fetch PR diff via gh pr view\",\n \"2. Analyze changes file by file for issues\",\n \"3. Check for bugs, security, and style problems\",\n \"4. Post review comments with gh pr review\",\n ],\n },\n {\n name: \"/test\",\n summary: \"Run and analyze test suites\",\n fullTokens: 290,\n content: [\n \"1. Detect test framework from package.json\",\n \"2. Run test suite and capture output\",\n \"3. Analyze failures and suggest fixes\",\n \"4. Re-run after applying fixes\",\n ],\n },\n {\n name: \"/deploy\",\n summary: \"Deploy application to target environment\",\n fullTokens: 350,\n content: [\n \"1. Verify all tests pass before deploy\",\n \"2. Build production bundle\",\n \"3. Push to deployment target via CI\",\n \"4. Verify health check on deployed URL\",\n ],\n },\n];\n\nconst TOKEN_STATES = [120, 120, 440, 440, 780, 780];\nconst MAX_TOKEN_DISPLAY = 1000;\n\nconst STEPS = [\n {\n title: \"Layer 1: Compact Summaries\",\n description:\n \"All skills are summarized in the system prompt. Compact, always present.\",\n },\n {\n title: \"Skill Invocation\",\n description:\n 'The model recognizes a skill invocation and triggers the Skill tool.',\n },\n {\n title: \"Layer 2: Full Injection\",\n description:\n \"The full skill instructions are injected as a tool_result, not into the system prompt.\",\n },\n {\n title: \"In Context Now\",\n description:\n \"The detailed instructions appear as if a tool returned them. The model follows them precisely.\",\n },\n {\n title: \"Stack Skills\",\n description:\n \"Multiple skills can be loaded. Only summaries are permanent; full content comes and goes.\",\n },\n {\n title: \"Two-Layer Architecture\",\n description:\n \"Layer 1: always present, tiny. Layer 2: loaded on demand, detailed. Elegant separation.\",\n },\n];\n\nexport default function SkillLoading({ title }: { title?: string }) {\n const {\n currentStep,\n totalSteps,\n next,\n prev,\n reset,\n isPlaying,\n toggleAutoPlay,\n } = useSteppedVisualization({ totalSteps: STEPS.length, autoPlayInterval: 2500 });\n\n const tokenCount = TOKEN_STATES[currentStep];\n const highlightedSkill = currentStep >= 1 && currentStep <= 3 ? 0 : currentStep >= 4 ? 1 : -1;\n const showFirstContent = currentStep >= 2;\n const showSecondContent = currentStep >= 4;\n const firstContentFaded = currentStep >= 5;\n\n return (\n \n \n {title || \"On-Demand Skill Loading\"}\n
\n\n \n \n {/* Main content area */}\n \n {/* System Prompt Block */}\n \n \n \n \n System Prompt\n \n \n always present\n \n \n \n \n # Available Skills\n \n \n {SKILLS.map((skill, i) => {\n const isHighlighted = i === highlightedSkill;\n return (\n \n \n {skill.name}\n \n {\" - \"}\n {skill.summary}\n \n );\n })}\n \n \n \n\n {/* User invocation indicator */}\n \n {currentStep === 1 && (\n \n \n User types:\n \n \n /commit\n \n \n )}\n {currentStep === 4 && (\n \n \n User types:\n \n \n /review-pr\n \n \n )}\n \n\n {/* Connecting arrow */}\n \n {(showFirstContent || showSecondContent) && (\n \n \n \n \n \n \n )}\n \n\n {/* Expanded Skill Content Blocks */}\n \n \n {showFirstContent && (\n \n \n \n \n \n \n SKILL.md: /commit\n \n \n \n tool_result\n \n \n \n {SKILLS[0].content.map((line, i) => (\n \n {line}\n \n ))}\n \n \n \n )}\n \n\n \n {showSecondContent && (\n \n \n \n \n \n \n SKILL.md: /review-pr\n \n \n \n tool_result\n \n \n \n {SKILLS[1].content.map((line, i) => (\n \n {line}\n \n ))}\n \n \n \n )}\n \n \n\n {/* Mechanism annotation on step 3 */}\n \n {currentStep === 3 && (\n \n The Skill tool returns content as a tool_result message.\n The model sees it in context and follows the instructions.\n No system prompt bloat.\n \n )}\n \n\n {/* Final overview label on step 5 */}\n \n {currentStep === 5 && (\n \n \n \n LAYER 1\n \n \n Always present, ~120 tokens\n \n \n \n \n LAYER 2\n \n \n On demand, ~300-500 tokens each\n \n \n \n )}\n \n \n\n {/* Token Gauge (vertical bar on the right) */}\n \n \n Tokens\n \n \n 600\n ? \"bg-amber-500\"\n : tokenCount > 300\n ? \"bg-blue-500\"\n : \"bg-emerald-500\"\n }`}\n />\n \n \n {tokenCount}\n \n \n \n\n {/* Step Controls */}\n \n \n );\n}\n"
},
{
"path": "web/src/components/visualizations/s06-context-compact.tsx",
"content": "\"use client\";\n\nimport { useMemo } from \"react\";\nimport { motion, AnimatePresence } from \"framer-motion\";\nimport { useSteppedVisualization } from \"@/hooks/useSteppedVisualization\";\nimport { StepControls } from \"@/components/visualizations/shared/step-controls\";\n\ntype BlockType = \"user\" | \"assistant\" | \"tool_result\";\n\ninterface ContextBlock {\n id: string;\n type: BlockType;\n label: string;\n tokens: number;\n}\n\nconst BLOCK_COLORS: Record = {\n user: \"bg-blue-500\",\n assistant: \"bg-zinc-500 dark:bg-zinc-600\",\n tool_result: \"bg-emerald-500\",\n};\n\nconst BLOCK_LABELS: Record = {\n user: \"USR\",\n assistant: \"AST\",\n tool_result: \"TRL\",\n};\n\nfunction generateBlocks(count: number, seed: number): ContextBlock[] {\n const types: BlockType[] = [\"user\", \"assistant\", \"tool_result\"];\n const blocks: ContextBlock[] = [];\n for (let i = 0; i < count; i++) {\n const typeIndex = (i + seed) % 3;\n const type = types[typeIndex];\n const tokens = type === \"tool_result\" ? 4000 + (i % 3) * 1000 : 1500 + (i % 4) * 500;\n blocks.push({\n id: `b-${seed}-${i}`,\n type,\n label: `${BLOCK_LABELS[type]} ${i + 1}`,\n tokens,\n });\n }\n return blocks;\n}\n\nconst MAX_TOKENS = 100000;\nconst WINDOW_HEIGHT = 350;\n\ninterface StepState {\n blocks: { id: string; type: BlockType; label: string; heightPx: number; compressed?: boolean }[];\n tokenCount: number;\n fillPercent: number;\n compressionLabel: string | null;\n}\n\nfunction computeStepState(step: number): StepState {\n switch (step) {\n case 0: {\n const raw = generateBlocks(8, 0);\n const tokenCount = 30000;\n const totalRawTokens = raw.reduce((a, b) => a + b.tokens, 0);\n const blocks = raw.map((b) => ({\n ...b,\n heightPx: Math.max(16, (b.tokens / totalRawTokens) * WINDOW_HEIGHT * 0.3),\n }));\n return { blocks, tokenCount, fillPercent: 30, compressionLabel: null };\n }\n case 1: {\n const raw = generateBlocks(16, 0);\n const tokenCount = 60000;\n const totalRawTokens = raw.reduce((a, b) => a + b.tokens, 0);\n const blocks = raw.map((b) => ({\n ...b,\n heightPx: Math.max(12, (b.tokens / totalRawTokens) * WINDOW_HEIGHT * 0.6),\n }));\n return { blocks, tokenCount, fillPercent: 60, compressionLabel: null };\n }\n case 2: {\n const raw = generateBlocks(20, 0);\n const tokenCount = 80000;\n const totalRawTokens = raw.reduce((a, b) => a + b.tokens, 0);\n const blocks = raw.map((b) => ({\n ...b,\n heightPx: Math.max(10, (b.tokens / totalRawTokens) * WINDOW_HEIGHT * 0.8),\n }));\n return { blocks, tokenCount, fillPercent: 80, compressionLabel: null };\n }\n case 3: {\n const raw = generateBlocks(20, 0);\n const tokenCount = 60000;\n const totalRawTokens = raw.reduce((a, b) => a + b.tokens, 0);\n const blocks = raw.map((b) => ({\n ...b,\n heightPx:\n b.type === \"tool_result\"\n ? 6\n : Math.max(12, (b.tokens / totalRawTokens) * WINDOW_HEIGHT * 0.6),\n compressed: b.type === \"tool_result\",\n }));\n return {\n blocks,\n tokenCount,\n fillPercent: 60,\n compressionLabel: \"MICRO-COMPACT\",\n };\n }\n case 4: {\n const raw = generateBlocks(24, 1);\n const tokenCount = 85000;\n const totalRawTokens = raw.reduce((a, b) => a + b.tokens, 0);\n const blocks = raw.map((b) => ({\n ...b,\n heightPx: Math.max(10, (b.tokens / totalRawTokens) * WINDOW_HEIGHT * 0.85),\n }));\n return { blocks, tokenCount, fillPercent: 85, compressionLabel: null };\n }\n case 5: {\n const tokenCount = 25000;\n const summaryBlock = {\n id: \"auto-summary\",\n type: \"assistant\" as BlockType,\n label: \"SUMMARY\",\n heightPx: 40,\n compressed: false,\n };\n const recentBlocks = generateBlocks(4, 2).map((b) => ({\n ...b,\n heightPx: 20,\n }));\n return {\n blocks: [summaryBlock, ...recentBlocks],\n tokenCount,\n fillPercent: 25,\n compressionLabel: \"AUTO-COMPACT\",\n };\n }\n case 6: {\n const tokenCount = 8000;\n const compactBlock = {\n id: \"compact-summary\",\n type: \"assistant\" as BlockType,\n label: \"COMPACT SUMMARY\",\n heightPx: 24,\n compressed: false,\n };\n return {\n blocks: [compactBlock],\n tokenCount,\n fillPercent: 8,\n compressionLabel: \"/compact\",\n };\n }\n default:\n return { blocks: [], tokenCount: 0, fillPercent: 0, compressionLabel: null };\n }\n}\n\nconst STEPS = [\n {\n title: \"Growing Context\",\n description:\n \"The context window holds the conversation. Each API call adds more messages.\",\n },\n {\n title: \"Context Growing\",\n description:\n \"As the agent works, messages accumulate. The context window fills up.\",\n },\n {\n title: \"Approaching Limit\",\n description:\n \"Old tool_results are the biggest consumers. Micro-compact targets these first.\",\n },\n {\n title: \"Stage 1: Micro-Compact\",\n description:\n \"Replace old tool_results with short summaries. Automatic, transparent to the model.\",\n },\n {\n title: \"Still Growing\",\n description:\n \"Work continues. Context grows again toward the threshold...\",\n },\n {\n title: \"Stage 2: Auto-Compact\",\n description:\n \"Entire conversation summarized into a compact block. Triggered at token threshold.\",\n },\n {\n title: \"Stage 3: /compact\",\n description:\n \"User-triggered, most aggressive. Three layers of strategic forgetting enable infinite sessions.\",\n },\n];\n\nexport default function ContextCompact({ title }: { title?: string }) {\n const {\n currentStep,\n totalSteps,\n next,\n prev,\n reset,\n isPlaying,\n toggleAutoPlay,\n } = useSteppedVisualization({ totalSteps: STEPS.length, autoPlayInterval: 2500 });\n\n const state = useMemo(() => computeStepState(currentStep), [currentStep]);\n\n const fillColor =\n state.fillPercent > 75\n ? \"bg-red-500\"\n : state.fillPercent > 45\n ? \"bg-amber-500\"\n : \"bg-emerald-500\";\n\n const tokenDisplay = `${(state.tokenCount / 1000).toFixed(0)}K`;\n\n return (\n \n \n {title || \"Three-Layer Context Compression\"}\n
\n\n \n \n {/* Token Window (tall vertical bar on the left) */}\n \n \n Context Window\n \n \n {/* Blocks stacked from bottom up */}\n \n \n {state.blocks.map((block) => (\n \n {block.heightPx >= 14 && (\n \n {block.label}\n \n )}\n \n ))}\n \n \n\n {/* Fill level line */}\n \n \n {state.fillPercent}%\n \n \n \n\n {/* Token count */}\n \n {tokenDisplay}\n \n \n / 100K\n \n \n\n {/* Right side: state display and compression stage */}\n \n {/* Top: horizontal token bar */}\n \n \n \n Token usage\n \n \n {state.tokenCount.toLocaleString()} / {MAX_TOKENS.toLocaleString()}\n \n \n \n \n {currentStep === 2 && (\n \n \n tool_results are the largest blocks\n \n \n File contents, command outputs, search results -- each one is thousands of tokens.\n \n \n )}\n \n\n {/* Compression stage label */}\n \n {state.compressionLabel && (\n \n \n \n {state.compressionLabel}\n \n \n {currentStep === 3 && \"Old tool_results shrunk to tiny summaries\"}\n {currentStep === 5 && \"Full conversation compressed to summary block\"}\n {currentStep === 6 && \"Most aggressive compression -- near-empty context\"}\n \n \n \n )}\n \n\n {/* Three stages overview on final step */}\n {currentStep === 6 && (\n \n \n \n \n Stage 1: Micro -- shrink old tool_results\n \n \n automatic\n \n \n \n \n \n Stage 2: Auto -- summarize entire conversation\n \n \n at threshold\n \n \n \n \n \n Stage 3: /compact -- user-triggered, deepest compression\n \n \n manual\n \n \n \n )}\n \n \n\n {/* Step Controls */}\n \n \n );\n}\n"
},
{
"path": "web/src/components/visualizations/s07-task-system.tsx",
"content": "\"use client\";\n\nimport { useMemo } from \"react\";\nimport { motion } from \"framer-motion\";\nimport { useSteppedVisualization } from \"@/hooks/useSteppedVisualization\";\nimport { StepControls } from \"@/components/visualizations/shared/step-controls\";\nimport { useDarkMode, useSvgPalette } from \"@/hooks/useDarkMode\";\n\ntype TaskStatus = \"pending\" | \"in_progress\" | \"completed\" | \"blocked\";\n\ninterface TaskNode {\n id: string;\n label: string;\n x: number;\n y: number;\n deps: string[];\n}\n\ninterface StepInfo {\n title: string;\n description: string;\n}\n\nconst TASKS: TaskNode[] = [\n { id: \"T1\", label: \"T1: Setup DB\", x: 80, y: 160, deps: [] },\n { id: \"T2\", label: \"T2: API routes\", x: 280, y: 80, deps: [\"T1\"] },\n { id: \"T3\", label: \"T3: Auth module\", x: 280, y: 240, deps: [\"T1\"] },\n { id: \"T4\", label: \"T4: Integration\", x: 480, y: 160, deps: [\"T2\", \"T3\"] },\n { id: \"T5\", label: \"T5: Deploy\", x: 650, y: 160, deps: [\"T4\"] },\n];\n\nconst NODE_W = 140;\nconst NODE_H = 50;\n\nconst STEP_INFO: StepInfo[] = [\n {\n title: \"File-Based Tasks\",\n description:\n \"Tasks are stored in JSON files on disk. They survive context compaction -- unlike in-memory state.\",\n },\n {\n title: \"Start T1\",\n description:\n \"Tasks without dependencies can start immediately. T1 has no blockers.\",\n },\n {\n title: \"T1 Complete\",\n description: \"Completing T1 unblocks its dependents: T2 and T3.\",\n },\n {\n title: \"Parallel Work\",\n description:\n \"T2 and T3 have no dependency on each other. Both can run simultaneously.\",\n },\n {\n title: \"Partial Unblock\",\n description:\n \"T4 depends on BOTH T2 and T3. It waits for all blockers to complete.\",\n },\n {\n title: \"Fully Unblocked\",\n description: \"All blockers resolved. T4 can now proceed.\",\n },\n {\n title: \"Graph Resolved\",\n description:\n \"The entire dependency graph is resolved. File-based persistence means this works across context compressions.\",\n },\n];\n\nfunction getTaskStatus(taskId: string, step: number): TaskStatus {\n const statusMap: Record = {\n T1: [\n \"pending\",\n \"in_progress\",\n \"completed\",\n \"completed\",\n \"completed\",\n \"completed\",\n \"completed\",\n ],\n T2: [\n \"pending\",\n \"pending\",\n \"pending\",\n \"in_progress\",\n \"completed\",\n \"completed\",\n \"completed\",\n ],\n T3: [\n \"pending\",\n \"pending\",\n \"pending\",\n \"in_progress\",\n \"in_progress\",\n \"completed\",\n \"completed\",\n ],\n T4: [\n \"pending\",\n \"pending\",\n \"pending\",\n \"pending\",\n \"blocked\",\n \"in_progress\",\n \"completed\",\n ],\n T5: [\n \"pending\",\n \"pending\",\n \"pending\",\n \"pending\",\n \"pending\",\n \"pending\",\n \"completed\",\n ],\n };\n return statusMap[taskId]?.[step] ?? \"pending\";\n}\n\nfunction isEdgeActive(fromId: string, toId: string, step: number): boolean {\n const fromStatus = getTaskStatus(fromId, step);\n const toStatus = getTaskStatus(toId, step);\n return (\n fromStatus === \"completed\" &&\n (toStatus === \"in_progress\" || toStatus === \"completed\")\n );\n}\n\nfunction getStatusColor(status: TaskStatus) {\n switch (status) {\n case \"pending\":\n return {\n fill: \"#e2e8f0\",\n darkFill: \"#27272a\",\n stroke: \"#cbd5e1\",\n darkStroke: \"#3f3f46\",\n text: \"#475569\",\n darkText: \"#d4d4d8\",\n };\n case \"in_progress\":\n return {\n fill: \"#fef3c7\",\n darkFill: \"#451a0340\",\n stroke: \"#f59e0b\",\n darkStroke: \"#d97706\",\n text: \"#b45309\",\n darkText: \"#fbbf24\",\n };\n case \"completed\":\n return {\n fill: \"#d1fae5\",\n darkFill: \"#06402740\",\n stroke: \"#10b981\",\n darkStroke: \"#059669\",\n text: \"#047857\",\n darkText: \"#34d399\",\n };\n case \"blocked\":\n return {\n fill: \"#fecaca\",\n darkFill: \"#45050540\",\n stroke: \"#ef4444\",\n darkStroke: \"#dc2626\",\n text: \"#dc2626\",\n darkText: \"#f87171\",\n };\n }\n}\n\nfunction getStatusLabel(status: TaskStatus): string {\n switch (status) {\n case \"pending\":\n return \"pending\";\n case \"in_progress\":\n return \"in_progress\";\n case \"completed\":\n return \"done\";\n case \"blocked\":\n return \"blocked\";\n }\n}\n\nfunction buildCurvePath(\n x1: number,\n y1: number,\n x2: number,\n y2: number\n): string {\n const midX = (x1 + x2) / 2;\n return `M ${x1} ${y1} C ${midX} ${y1}, ${midX} ${y2}, ${x2} ${y2}`;\n}\n\nexport default function TaskSystem({ title }: { title?: string }) {\n const {\n currentStep,\n totalSteps,\n next,\n prev,\n reset,\n isPlaying,\n toggleAutoPlay,\n } = useSteppedVisualization({ totalSteps: 7, autoPlayInterval: 2500 });\n\n const isDark = useDarkMode();\n const palette = useSvgPalette();\n\n const edges = useMemo(() => {\n const result: {\n fromId: string;\n toId: string;\n x1: number;\n y1: number;\n x2: number;\n y2: number;\n }[] = [];\n for (const task of TASKS) {\n for (const depId of task.deps) {\n const dep = TASKS.find((t) => t.id === depId);\n if (!dep) continue;\n result.push({\n fromId: dep.id,\n toId: task.id,\n x1: dep.x + NODE_W,\n y1: dep.y + NODE_H / 2,\n x2: task.x,\n y2: task.y + NODE_H / 2,\n });\n }\n }\n return result;\n }, []);\n\n const stepInfo = STEP_INFO[currentStep];\n\n return (\n \n \n {title || \"Task Dependency Graph\"}\n
\n\n \n \n\n {/* File persistence indicator */}\n \n \n \n\n {/* Legend */}\n \n \n \n \n pending\n \n \n \n \n \n in_progress\n \n \n \n \n \n completed\n \n \n \n \n \n blocked\n \n \n \n \n\n \n );\n}\n"
},
{
"path": "web/src/components/visualizations/s08-background-tasks.tsx",
"content": "\"use client\";\n\nimport { motion, AnimatePresence } from \"framer-motion\";\nimport { useSteppedVisualization } from \"@/hooks/useSteppedVisualization\";\nimport { StepControls } from \"@/components/visualizations/shared/step-controls\";\nimport { useDarkMode, useSvgPalette } from \"@/hooks/useDarkMode\";\n\ninterface StepInfo {\n title: string;\n description: string;\n}\n\nconst STEP_INFO: StepInfo[] = [\n {\n title: \"Three Lanes\",\n description:\n \"The agent has a main thread and can spawn daemon background threads for parallel work.\",\n },\n {\n title: \"Main Thread Working\",\n description:\n \"The main agent loop runs as usual, processing user requests.\",\n },\n {\n title: \"Spawn Background\",\n description:\n \"Background tasks run as daemon threads. The main loop doesn't wait for them.\",\n },\n {\n title: \"Multiple Backgrounds\",\n description: \"Multiple background tasks can run concurrently.\",\n },\n {\n title: \"Task Completes\",\n description:\n \"Background task finishes. Its result goes to the notification queue.\",\n },\n {\n title: \"Queue Fills\",\n description:\n \"Results accumulate in the queue, invisible to the model during this turn.\",\n },\n {\n title: \"Drain Queue\",\n description:\n \"Just before the next LLM call, all queued notifications are injected as tool_results. Non-blocking, async.\",\n },\n];\n\nconst LANE_Y = {\n main: 60,\n bg1: 140,\n bg2: 220,\n} as const;\n\nconst LANE_HEIGHT = 44;\nconst TIMELINE_LEFT = 160;\nconst TIMELINE_RIGHT = 720;\nconst TIMELINE_WIDTH = TIMELINE_RIGHT - TIMELINE_LEFT;\n\nconst QUEUE_Y = 300;\n\ninterface WorkBlock {\n lane: \"main\" | \"bg1\" | \"bg2\";\n startFraction: number;\n endFraction: number;\n color: string;\n label?: string;\n appearsAtStep: number;\n completesAtStep?: number;\n}\n\nconst WORK_BLOCKS: WorkBlock[] = [\n {\n lane: \"main\",\n startFraction: 0,\n endFraction: 1,\n color: \"#8b5cf6\",\n label: \"Main agent loop\",\n appearsAtStep: 1,\n },\n {\n lane: \"bg1\",\n startFraction: 0.18,\n endFraction: 0.75,\n color: \"#10b981\",\n label: \"Run tests\",\n appearsAtStep: 2,\n completesAtStep: 5,\n },\n {\n lane: \"bg2\",\n startFraction: 0.35,\n endFraction: 0.58,\n color: \"#3b82f6\",\n label: \"Lint code\",\n appearsAtStep: 3,\n completesAtStep: 4,\n },\n];\n\ninterface ForkArrow {\n fromFraction: number;\n toLane: \"bg1\" | \"bg2\";\n appearsAtStep: number;\n}\n\nconst FORK_ARROWS: ForkArrow[] = [\n { fromFraction: 0.18, toLane: \"bg1\", appearsAtStep: 2 },\n { fromFraction: 0.35, toLane: \"bg2\", appearsAtStep: 3 },\n];\n\ninterface QueueCard {\n id: string;\n label: string;\n appearsAtStep: number;\n drainsAtStep: number;\n}\n\nconst QUEUE_CARDS: QueueCard[] = [\n {\n id: \"lint-result\",\n label: \"Lint: 0 errors\",\n appearsAtStep: 4,\n drainsAtStep: 6,\n },\n {\n id: \"test-result\",\n label: \"Tests: 42 passed\",\n appearsAtStep: 5,\n drainsAtStep: 6,\n },\n];\n\nfunction fractionToX(fraction: number): number {\n return TIMELINE_LEFT + fraction * TIMELINE_WIDTH;\n}\n\nfunction getBlockEndFraction(block: WorkBlock, step: number): number {\n if (step < block.appearsAtStep) return block.startFraction;\n if (block.completesAtStep !== undefined && step >= block.completesAtStep) {\n return block.endFraction;\n }\n const growthSteps = (block.completesAtStep ?? 6) - block.appearsAtStep;\n const stepsElapsed = step - block.appearsAtStep;\n const progress = Math.min(stepsElapsed / growthSteps, 1);\n const range = block.endFraction - block.startFraction;\n return block.startFraction + range * progress;\n}\n\nexport default function BackgroundTasks({ title }: { title?: string }) {\n const {\n currentStep,\n totalSteps,\n next,\n prev,\n reset,\n isPlaying,\n toggleAutoPlay,\n } = useSteppedVisualization({ totalSteps: 7, autoPlayInterval: 2500 });\n\n const isDark = useDarkMode();\n const palette = useSvgPalette();\n\n const stepInfo = STEP_INFO[currentStep];\n\n const llmCallFraction = 0.82;\n const showLlmMarker = currentStep >= 5;\n\n return (\n \n \n {title || \"Background Task Lanes\"}\n
\n\n \n \n\n {/* Legend */}\n \n \n \n \n Main thread\n \n \n \n \n \n Background 1\n \n \n \n \n \n Background 2\n \n \n \n \n \n LLM boundary\n \n \n \n \n\n \n );\n}\n"
},
{
"path": "web/src/components/visualizations/s09-agent-teams.tsx",
"content": "\"use client\";\n\nimport { motion, AnimatePresence } from \"framer-motion\";\nimport { useSteppedVisualization } from \"@/hooks/useSteppedVisualization\";\nimport { StepControls } from \"@/components/visualizations/shared/step-controls\";\nimport { useSvgPalette } from \"@/hooks/useDarkMode\";\n\n// -- Layout constants --\nconst SVG_W = 560;\nconst SVG_H = 340;\nconst AGENT_R = 40;\n\n// Agent positions: inverted triangle (Lead top-center, Coder bottom-left, Reviewer bottom-right)\nconst AGENTS = [\n { id: \"lead\", label: \"Lead\", cx: SVG_W / 2, cy: 70, inbox: \"lead.jsonl\" },\n { id: \"coder\", label: \"Coder\", cx: 140, cy: 230, inbox: \"coder.jsonl\" },\n { id: \"reviewer\", label: \"Reviewer\", cx: SVG_W - 140, cy: 230, inbox: \"reviewer.jsonl\" },\n] as const;\n\n// Inbox tray dimensions, positioned below each agent circle\nconst TRAY_W = 72;\nconst TRAY_H = 22;\nconst TRAY_OFFSET_Y = AGENT_R + 14;\n\n// Message block dimensions\nconst MSG_W = 60;\nconst MSG_H = 20;\n\nfunction agentById(id: string) {\n return AGENTS.find((a) => a.id === id)!;\n}\n\nfunction trayCenter(id: string) {\n const a = agentById(id);\n return { x: a.cx, y: a.cy + TRAY_OFFSET_Y + TRAY_H / 2 };\n}\n\n// Step configuration\nconst STEPS = [\n { title: \"The Team\", desc: \"Teams use a leader-worker pattern. Each teammate has a file-based mailbox inbox.\" },\n { title: \"Lead Assigns Work\", desc: \"Communication is async: write a message to the recipient's .jsonl inbox file.\" },\n { title: \"Read Inbox\", desc: \"Teammates poll their inbox before each LLM call. New messages become context.\" },\n { title: \"Independent Work\", desc: \"Each teammate runs its own agent loop independently.\" },\n { title: \"Pass Result\", desc: \"Results flow through the same mailbox mechanism. All communication is via files.\" },\n { title: \"Feedback Loop\", desc: \"The mailbox pattern supports any communication topology: linear, broadcast, round-robin.\" },\n { title: \"File-Based Coordination\", desc: \"No shared memory, no locks. All coordination through append-only files. Simple, robust, debuggable.\" },\n];\n\n// Helper: determine which agent glows at each step\nfunction agentGlows(agentId: string, step: number): boolean {\n if (step === 1 && agentId === \"lead\") return true;\n if (step === 2 && agentId === \"coder\") return true;\n if (step === 3 && agentId === \"coder\") return true;\n if (step === 4 && agentId === \"coder\") return true;\n if (step === 5 && agentId === \"reviewer\") return true;\n return false;\n}\n\n// Helper: determine which inbox tray has a message sitting in it\nfunction trayHasMessage(agentId: string, step: number): boolean {\n if (step === 2 && agentId === \"coder\") return true;\n if (step === 4 && agentId === \"reviewer\") return false;\n if (step === 5 && agentId === \"reviewer\") return true;\n return false;\n}\n\n// Animated message that travels from one point to another\nfunction TravelingMessage({\n fromX,\n fromY,\n toX,\n toY,\n label,\n delay = 0,\n}: {\n fromX: number;\n fromY: number;\n toX: number;\n toY: number;\n label: string;\n delay?: number;\n}) {\n return (\n \n \n {label}\n \n \n );\n}\n\n// Faded trace line between two agents\nfunction TraceLine({ from, to, strokeColor }: { from: string; to: string; strokeColor: string }) {\n const f = trayCenter(from);\n const t = trayCenter(to);\n return (\n \n \n {title || \"Agent Team Mailboxes\"}\n
\n \n \n {/* SVG visualization */}\n \n \n \n \n\n {/* Step controls */}\n \n \n );\n}\n"
},
{
"path": "web/src/components/visualizations/s10-team-protocols.tsx",
"content": "\"use client\";\n\nimport { useState } from \"react\";\nimport { motion, AnimatePresence } from \"framer-motion\";\nimport { useSteppedVisualization } from \"@/hooks/useSteppedVisualization\";\nimport { StepControls } from \"@/components/visualizations/shared/step-controls\";\nimport { useSvgPalette } from \"@/hooks/useDarkMode\";\n\ntype Protocol = \"shutdown\" | \"plan\";\n\n// -- Layout constants for the sequence diagram --\nconst SVG_W = 560;\nconst SVG_H = 360;\nconst LIFELINE_LEFT_X = 140;\nconst LIFELINE_RIGHT_X = 420;\nconst LIFELINE_TOP = 60;\nconst LIFELINE_BOTTOM = 330;\nconst ACTIVATION_W = 12;\nconst ARROW_Y_START = 110;\nconst ARROW_Y_GAP = 70;\n\n// Request ID shown on message tags\nconst REQUEST_ID = \"req_abc\";\n\n// -- Shutdown protocol step definitions --\nconst SHUTDOWN_STEPS = [\n { title: \"Structured Protocols\", desc: \"Protocols define structured message exchanges with correlated request IDs.\" },\n { title: \"Shutdown Request\", desc: \"The leader initiates shutdown. The request_id links the request to its response.\" },\n { title: \"Teammate Decides\", desc: \"The teammate can accept or reject. It's not a forced kill -- it's a polite request.\" },\n { title: \"Approved\", desc: \"Same request_id in the response. Teammate exits cleanly.\" },\n];\n\n// -- Plan approval protocol step definitions --\nconst PLAN_STEPS = [\n { title: \"Plan Approval\", desc: \"Teammates in plan_mode must get approval before implementing changes.\" },\n { title: \"Submit Plan\", desc: \"The teammate designs a plan and sends it to the leader for review.\" },\n { title: \"Leader Reviews\", desc: \"Leader reviews and approves or rejects with feedback. Same request-response pattern.\" },\n];\n\n// Horizontal arrow between lifelines\nfunction SequenceArrow({\n y,\n direction,\n label,\n tagLabel,\n color,\n tagBg,\n tagStroke,\n tagText,\n}: {\n y: number;\n direction: \"right\" | \"left\";\n label: string;\n tagLabel?: string;\n color: string;\n tagBg?: string;\n tagStroke?: string;\n tagText?: string;\n}) {\n const fromX = direction === \"right\" ? LIFELINE_LEFT_X + ACTIVATION_W / 2 : LIFELINE_RIGHT_X - ACTIVATION_W / 2;\n const toX = direction === \"right\" ? LIFELINE_RIGHT_X - ACTIVATION_W / 2 : LIFELINE_LEFT_X + ACTIVATION_W / 2;\n const arrowTip = direction === \"right\" ? toX - 6 : toX + 6;\n const labelX = (fromX + toX) / 2;\n\n return (\n \n {/* Arrow line */}\n \n {label}\n \n {/* Request ID tag */}\n {tagLabel && (\n \n \n {tagLabel}\n \n \n )}\n \n );\n}\n\n// Decision diamond on a lifeline\nfunction DecisionBox({ x, y }: { x: number; y: number }) {\n const size = 14;\n return (\n \n \n ?\n \n \n approve\n \n \n reject\n \n \n );\n}\n\n// Activation bar on a lifeline\nfunction ActivationBar({\n x,\n yStart,\n yEnd,\n color,\n}: {\n x: number;\n yStart: number;\n yEnd: number;\n color: string;\n}) {\n return (\n (\"shutdown\");\n\n const totalSteps = protocol === \"shutdown\" ? SHUTDOWN_STEPS.length : PLAN_STEPS.length;\n const steps = protocol === \"shutdown\" ? SHUTDOWN_STEPS : PLAN_STEPS;\n\n const vis = useSteppedVisualization({ totalSteps, autoPlayInterval: 2500 });\n const step = vis.currentStep;\n const palette = useSvgPalette();\n\n const switchProtocol = (p: Protocol) => {\n setProtocol(p);\n vis.reset();\n };\n\n const leftLabel = protocol === \"shutdown\" ? \"Leader\" : \"Leader\";\n const rightLabel = protocol === \"shutdown\" ? \"Teammate\" : \"Teammate\";\n\n return (\n \n \n {title || \"FSM Team Protocols\"}\n
\n \n {/* Protocol toggle */}\n \n switchProtocol(\"shutdown\")}\n className={`rounded-md px-4 py-1.5 text-xs font-medium transition-colors ${\n protocol === \"shutdown\"\n ? \"bg-blue-500 text-white\"\n : \"bg-zinc-100 text-zinc-600 hover:bg-zinc-200 dark:bg-zinc-800 dark:text-zinc-400 dark:hover:bg-zinc-700\"\n }`}\n >\n Shutdown Protocol\n \n switchProtocol(\"plan\")}\n className={`rounded-md px-4 py-1.5 text-xs font-medium transition-colors ${\n protocol === \"plan\"\n ? \"bg-emerald-500 text-white\"\n : \"bg-zinc-100 text-zinc-600 hover:bg-zinc-200 dark:bg-zinc-800 dark:text-zinc-400 dark:hover:bg-zinc-700\"\n }`}\n >\n Plan Approval Protocol\n \n \n\n {/* Sequence diagram SVG */}\n \n\n {/* Step controls */}\n \n \n );\n}\n"
},
{
"path": "web/src/components/visualizations/s11-autonomous-agents.tsx",
"content": "\"use client\";\n\nimport { motion } from \"framer-motion\";\nimport { useSteppedVisualization } from \"@/hooks/useSteppedVisualization\";\nimport { StepControls } from \"@/components/visualizations/shared/step-controls\";\nimport { useSvgPalette } from \"@/hooks/useDarkMode\";\n\n// -- FSM states and their layout positions (diamond: idle top, poll right, claim bottom, work left) --\ntype Phase = \"idle\" | \"poll\" | \"claim\" | \"work\";\n\nconst FSM_CX = 110;\nconst FSM_CY = 110;\nconst FSM_R = 65;\nconst FSM_STATE_R = 22;\n\nconst FSM_STATES: { id: Phase; label: string; angle: number }[] = [\n { id: \"idle\", label: \"idle\", angle: -Math.PI / 2 },\n { id: \"poll\", label: \"poll\", angle: 0 },\n { id: \"claim\", label: \"claim\", angle: Math.PI / 2 },\n { id: \"work\", label: \"work\", angle: Math.PI },\n];\n\nconst FSM_TRANSITIONS: { from: Phase; to: Phase }[] = [\n { from: \"idle\", to: \"poll\" },\n { from: \"poll\", to: \"claim\" },\n { from: \"claim\", to: \"work\" },\n { from: \"work\", to: \"idle\" },\n];\n\nfunction fsmPos(angle: number) {\n return { x: FSM_CX + FSM_R * Math.cos(angle), y: FSM_CY + FSM_R * Math.sin(angle) };\n}\n\nconst PHASE_COLORS: Record = {\n idle: \"#a1a1aa\",\n poll: \"#f59e0b\",\n claim: \"#3b82f6\",\n work: \"#10b981\",\n};\n\n// -- Task board data --\ninterface TaskRow {\n id: string;\n name: string;\n status: \"unclaimed\" | \"active\" | \"complete\";\n owner: string;\n}\n\nconst INITIAL_TASKS: TaskRow[] = [\n { id: \"T1\", name: \"Fix auth bug\", status: \"unclaimed\", owner: \"-\" },\n { id: \"T2\", name: \"Add rate limiter\", status: \"unclaimed\", owner: \"-\" },\n { id: \"T3\", name: \"Write tests\", status: \"unclaimed\", owner: \"-\" },\n { id: \"T4\", name: \"Update API docs\", status: \"unclaimed\", owner: \"-\" },\n];\n\n// Agent positions around the task board (left panel)\nconst BOARD_CX = 140;\nconst BOARD_CY = 90;\nconst AGENT_ORBIT = 85;\nconst AGENT_R = 20;\n\nconst AGENT_ANGLES = [-Math.PI / 2, Math.PI / 6, (5 * Math.PI) / 6];\n\nfunction agentPos(index: number) {\n const angle = AGENT_ANGLES[index];\n return { x: BOARD_CX + AGENT_ORBIT * Math.cos(angle), y: BOARD_CY + AGENT_ORBIT * Math.sin(angle) };\n}\n\n// -- Step definitions --\nconst STEPS = [\n { title: \"Self-Governing Agents\", desc: \"Autonomous agents need no coordinator. They govern themselves with an idle-poll-claim-work cycle.\" },\n { title: \"Idle Timer\", desc: \"Each idle agent counts rounds. A timeout triggers self-directed task polling.\" },\n { title: \"Poll Task Board\", desc: \"Timeout! The agent reads the task board looking for unclaimed work.\" },\n { title: \"Claim Task\", desc: \"The agent writes its name to the task record. Atomic, no conflicts.\" },\n { title: \"Work\", desc: \"The agent works on the claimed task using its own agent loop.\" },\n { title: \"Independent Polling\", desc: \"Multiple agents poll and claim independently. No central coordinator needed.\" },\n { title: \"Complete & Reset\", desc: \"Task done. Agent returns to idle. The cycle repeats.\" },\n { title: \"Self-Organization\", desc: \"Three agents, zero coordination overhead. Polling + timeout = emergent organization.\" },\n];\n\n// Per-step state for each agent\ninterface AgentState {\n phase: Phase;\n timerFill: number;\n color: string;\n taskClaim: string | null;\n}\n\nfunction getAgentStates(step: number): AgentState[] {\n const idle: AgentState = { phase: \"idle\", timerFill: 0, color: PHASE_COLORS.idle, taskClaim: null };\n\n switch (step) {\n case 0:\n return [\n { ...idle },\n { ...idle },\n { ...idle },\n ];\n case 1:\n return [\n { phase: \"idle\", timerFill: 0.6, color: PHASE_COLORS.idle, taskClaim: null },\n { ...idle },\n { ...idle },\n ];\n case 2:\n return [\n { phase: \"poll\", timerFill: 1.0, color: PHASE_COLORS.poll, taskClaim: null },\n { ...idle },\n { ...idle },\n ];\n case 3:\n return [\n { phase: \"claim\", timerFill: 0, color: PHASE_COLORS.claim, taskClaim: \"T1\" },\n { ...idle },\n { ...idle },\n ];\n case 4:\n return [\n { phase: \"work\", timerFill: 0, color: PHASE_COLORS.work, taskClaim: \"T1\" },\n { ...idle },\n { ...idle },\n ];\n case 5:\n return [\n { phase: \"work\", timerFill: 0, color: PHASE_COLORS.work, taskClaim: \"T1\" },\n { phase: \"claim\", timerFill: 0, color: PHASE_COLORS.claim, taskClaim: \"T2\" },\n { ...idle },\n ];\n case 6:\n return [\n { phase: \"idle\", timerFill: 0, color: PHASE_COLORS.idle, taskClaim: null },\n { phase: \"work\", timerFill: 0, color: PHASE_COLORS.work, taskClaim: \"T2\" },\n { ...idle },\n ];\n case 7:\n return [\n { phase: \"idle\", timerFill: 0, color: PHASE_COLORS.idle, taskClaim: null },\n { phase: \"work\", timerFill: 0, color: PHASE_COLORS.work, taskClaim: \"T2\" },\n { phase: \"claim\", timerFill: 0, color: PHASE_COLORS.claim, taskClaim: \"T3\" },\n ];\n default:\n return [{ ...idle }, { ...idle }, { ...idle }];\n }\n}\n\nfunction getTaskStates(step: number): TaskRow[] {\n const tasks = INITIAL_TASKS.map((t) => ({ ...t }));\n if (step >= 3) { tasks[0].status = \"active\"; tasks[0].owner = \"A\"; }\n if (step >= 5) { tasks[1].status = \"active\"; tasks[1].owner = \"B\"; }\n if (step >= 6) { tasks[0].status = \"complete\"; }\n if (step >= 7) { tasks[2].status = \"active\"; tasks[2].owner = \"C\"; }\n return tasks;\n}\n\nfunction getActivePhase(step: number): Phase {\n if (step <= 1) return \"idle\";\n if (step === 2) return \"poll\";\n if (step === 3) return \"claim\";\n if (step === 4 || step === 5) return \"work\";\n if (step === 6) return \"idle\";\n return \"claim\";\n}\n\n// Ring timer around an agent\nfunction TimerRing({ cx, cy, r, fill }: { cx: number; cy: number; r: number; fill: number }) {\n if (fill <= 0) return null;\n const circumference = 2 * Math.PI * (r + 4);\n const offset = circumference * (1 - fill);\n return (\n \n \n );\n}\n\nexport default function AutonomousAgents({ title }: { title?: string }) {\n const vis = useSteppedVisualization({ totalSteps: STEPS.length, autoPlayInterval: 2500 });\n const step = vis.currentStep;\n const palette = useSvgPalette();\n\n const agentStates = getAgentStates(step);\n const tasks = getTaskStates(step);\n const activePhase = getActivePhase(step);\n const agentNames = [\"A\", \"B\", \"C\"];\n\n return (\n \n \n {title || \"Autonomous Agent Cycle\"}\n
\n \n \n {/* Left panel: spatial view with agents and task board */}\n \n Spatial View\n \n\n {/* Task table below the spatial view */}\n \n \n \n \n Task \n Status \n Owner \n \n \n \n {tasks.map((task) => (\n \n {task.name} \n \n \n {task.status}\n \n \n {task.owner} \n \n ))}\n \n
\n \n \n\n {/* Right panel: FSM state machine diagram */}\n \n FSM Cycle\n \n\n {/* Legend */}\n \n {FSM_STATES.map((s) => (\n \n \n {s.label}\n \n ))}\n \n \n \n\n {/* Step controls */}\n \n \n );\n}\n"
},
{
"path": "web/src/components/visualizations/s12-worktree-task-isolation.tsx",
"content": "\"use client\";\n\nimport { motion } from \"framer-motion\";\nimport { useSteppedVisualization } from \"@/hooks/useSteppedVisualization\";\nimport { StepControls } from \"@/components/visualizations/shared/step-controls\";\n\ntype TaskStatus = \"pending\" | \"in_progress\" | \"completed\";\n\ninterface TaskRow {\n id: number;\n subject: string;\n status: TaskStatus;\n worktree: string;\n}\n\ninterface WorktreeRow {\n name: string;\n branch: string;\n task: string;\n state: \"none\" | \"active\" | \"kept\" | \"removed\";\n}\n\ninterface Lane {\n name: string;\n files: string[];\n highlight?: boolean;\n}\n\ninterface StepState {\n title: string;\n desc: string;\n tasks: TaskRow[];\n worktrees: WorktreeRow[];\n lanes: Lane[];\n op: string;\n}\n\nconst STEPS: StepState[] = [\n {\n title: \"Single Workspace Pain\",\n desc: \"Two tasks are active, but both edits would hit one directory and collide.\",\n op: \"task_create x2\",\n tasks: [\n { id: 1, subject: \"Auth refactor\", status: \"in_progress\", worktree: \"\" },\n { id: 2, subject: \"UI login polish\", status: \"in_progress\", worktree: \"\" },\n ],\n worktrees: [],\n lanes: [\n { name: \"main\", files: [\"auth/service.py\", \"ui/Login.tsx\"], highlight: true },\n { name: \"wt/auth-refactor\", files: [] },\n { name: \"wt/ui-login\", files: [] },\n ],\n },\n {\n title: \"Allocate Lane for Task 1\",\n desc: \"Create a worktree lane and associate it with task 1 for clear ownership.\",\n op: \"worktree_create(name='auth-refactor', task_id=1)\",\n tasks: [\n { id: 1, subject: \"Auth refactor\", status: \"in_progress\", worktree: \"auth-refactor\" },\n { id: 2, subject: \"UI login polish\", status: \"in_progress\", worktree: \"\" },\n ],\n worktrees: [\n { name: \"auth-refactor\", branch: \"wt/auth-refactor\", task: \"#1\", state: \"active\" },\n ],\n lanes: [\n { name: \"main\", files: [\"ui/Login.tsx\"] },\n { name: \"wt/auth-refactor\", files: [\"auth/service.py\"], highlight: true },\n { name: \"wt/ui-login\", files: [] },\n ],\n },\n {\n title: \"Allocate Lane for Task 2\",\n desc: \"Lane creation and task association can be separate. Here task 2 binds after lane creation.\",\n op: \"worktree_create(name='ui-login')\\ntask_bind_worktree(task_id=2, worktree='ui-login')\",\n tasks: [\n { id: 1, subject: \"Auth refactor\", status: \"in_progress\", worktree: \"auth-refactor\" },\n { id: 2, subject: \"UI login polish\", status: \"in_progress\", worktree: \"ui-login\" },\n ],\n worktrees: [\n { name: \"auth-refactor\", branch: \"wt/auth-refactor\", task: \"#1\", state: \"active\" },\n { name: \"ui-login\", branch: \"wt/ui-login\", task: \"#2\", state: \"active\" },\n ],\n lanes: [\n { name: \"main\", files: [] },\n { name: \"wt/auth-refactor\", files: [\"auth/service.py\"] },\n { name: \"wt/ui-login\", files: [\"ui/Login.tsx\"], highlight: true },\n ],\n },\n {\n title: \"Run Commands in Isolated Lanes\",\n desc: \"Each command routes by selected lane directory, not by the shared root.\",\n op: \"worktree_run('auth-refactor', 'pytest tests/auth -q')\",\n tasks: [\n { id: 1, subject: \"Auth refactor\", status: \"in_progress\", worktree: \"auth-refactor\" },\n { id: 2, subject: \"UI login polish\", status: \"in_progress\", worktree: \"ui-login\" },\n ],\n worktrees: [\n { name: \"auth-refactor\", branch: \"wt/auth-refactor\", task: \"#1\", state: \"active\" },\n { name: \"ui-login\", branch: \"wt/ui-login\", task: \"#2\", state: \"active\" },\n ],\n lanes: [\n { name: \"main\", files: [] },\n { name: \"wt/auth-refactor\", files: [\"auth/service.py\", \"tests/auth/test_login.py\"], highlight: true },\n { name: \"wt/ui-login\", files: [\"ui/Login.tsx\", \"ui/Login.css\"] },\n ],\n },\n {\n title: \"Keep One Lane, Close Another\",\n desc: \"Closeout can mix decisions: keep ui-login active for follow-up, remove auth-refactor and complete task 1.\",\n op: \"worktree_keep('ui-login')\\nworktree_remove('auth-refactor', complete_task=true)\\nworktree_events(limit=10)\",\n tasks: [\n { id: 1, subject: \"Auth refactor\", status: \"completed\", worktree: \"\" },\n { id: 2, subject: \"UI login polish\", status: \"in_progress\", worktree: \"ui-login\" },\n ],\n worktrees: [\n { name: \"auth-refactor\", branch: \"wt/auth-refactor\", task: \"#1\", state: \"removed\" },\n { name: \"ui-login\", branch: \"wt/ui-login\", task: \"#2\", state: \"kept\" },\n ],\n lanes: [\n { name: \"main\", files: [] },\n { name: \"wt/auth-refactor\", files: [] },\n { name: \"wt/ui-login\", files: [\"ui/Login.tsx\"], highlight: true },\n ],\n },\n {\n title: \"Isolation + Coordination + Events\",\n desc: \"The board tracks shared truth, worktree lanes isolate execution, and events provide auditable side-channel traces.\",\n op: \"task_list + worktree_list + worktree_events\",\n tasks: [\n { id: 1, subject: \"Auth refactor\", status: \"completed\", worktree: \"\" },\n { id: 2, subject: \"UI login polish\", status: \"in_progress\", worktree: \"ui-login\" },\n ],\n worktrees: [\n { name: \"auth-refactor\", branch: \"wt/auth-refactor\", task: \"#1\", state: \"removed\" },\n { name: \"ui-login\", branch: \"wt/ui-login\", task: \"#2\", state: \"kept\" },\n ],\n lanes: [\n { name: \"main\", files: [] },\n { name: \"wt/auth-refactor\", files: [] },\n { name: \"wt/ui-login\", files: [\"ui/Login.tsx\"], highlight: true },\n ],\n },\n];\n\nfunction statusClass(status: TaskStatus): string {\n if (status === \"completed\") return \"bg-emerald-100 text-emerald-700 dark:bg-emerald-900/30 dark:text-emerald-300\";\n if (status === \"in_progress\") return \"bg-amber-100 text-amber-700 dark:bg-amber-900/30 dark:text-amber-300\";\n return \"bg-zinc-100 text-zinc-700 dark:bg-zinc-800 dark:text-zinc-300\";\n}\n\nfunction worktreeClass(state: WorktreeRow[\"state\"]): string {\n if (state === \"active\") return \"border-emerald-300 bg-emerald-50 dark:border-emerald-800 dark:bg-emerald-900/20\";\n if (state === \"kept\") return \"border-sky-300 bg-sky-50 dark:border-sky-800 dark:bg-sky-900/20\";\n if (state === \"removed\") return \"border-zinc-200 bg-zinc-100 opacity-70 dark:border-zinc-700 dark:bg-zinc-800\";\n return \"border-zinc-200 bg-white dark:border-zinc-700 dark:bg-zinc-900\";\n}\n\nexport default function WorktreeTaskIsolation({ title }: { title?: string }) {\n const vis = useSteppedVisualization({ totalSteps: STEPS.length, autoPlayInterval: 2600 });\n const step = STEPS[vis.currentStep];\n\n return (\n \n \n {title || \"Worktree Task Isolation\"}\n
\n\n \n \n {step.op}\n \n\n \n \n \n Task Board (.tasks)\n \n \n {step.tasks.map((task) => (\n \n \n #{task.id}\n \n {task.status}\n \n \n {task.subject}\n \n worktree: {task.worktree || \"-\"}\n \n \n ))}\n \n \n\n \n \n Worktree Index (.worktrees/index.json)\n \n \n {step.worktrees.length === 0 && (\n \n no worktrees yet\n \n )}\n {step.worktrees.map((wt) => (\n \n {wt.name}\n {wt.branch}\n task: {wt.task}\n \n ))}\n \n \n\n \n \n Execution Lanes\n \n \n {step.lanes.map((lane) => (\n \n {lane.name}\n \n {lane.files.length === 0 ? (\n (no changes)\n ) : (\n lane.files.map((f) => {f})\n )}\n \n \n ))}\n \n \n \n\n \n {step.title}\n {step.desc}\n \n \n\n \n );\n}\n"
},
{
"path": "web/src/components/visualizations/shared/step-controls.tsx",
"content": "\"use client\";\n\nimport { Play, Pause, SkipBack, SkipForward, RotateCcw } from \"lucide-react\";\nimport { cn } from \"@/lib/utils\";\n\ninterface StepControlsProps {\n currentStep: number;\n totalSteps: number;\n onPrev: () => void;\n onNext: () => void;\n onReset: () => void;\n isPlaying: boolean;\n onToggleAutoPlay: () => void;\n stepTitle: string;\n stepDescription: string;\n className?: string;\n}\n\nexport function StepControls({\n currentStep,\n totalSteps,\n onPrev,\n onNext,\n onReset,\n isPlaying,\n onToggleAutoPlay,\n stepTitle,\n stepDescription,\n className,\n}: StepControlsProps) {\n return (\n \n {/* Annotation */}\n \n \n {stepTitle}\n \n \n {stepDescription}\n \n \n\n {/* Controls */}\n \n \n \n \n \n {isPlaying ? \n \n\n {/* Step indicator */}\n \n \n {Array.from({ length: totalSteps }, (_, i) => (\n = {\n s01: {\n nodes: [\n { id: \"start\", label: \"User Input\", type: \"start\", x: COL_CENTER, y: 30 },\n { id: \"llm\", label: \"LLM Call\", type: \"process\", x: COL_CENTER, y: 110 },\n { id: \"tool_check\", label: \"tool_use?\", type: \"decision\", x: COL_CENTER, y: 190 },\n { id: \"bash\", label: \"Execute Bash\", type: \"subprocess\", x: COL_LEFT, y: 280 },\n { id: \"append\", label: \"Append Result\", type: \"process\", x: COL_LEFT, y: 360 },\n { id: \"end\", label: \"Output\", type: \"end\", x: COL_RIGHT, y: 280 },\n ],\n edges: [\n { from: \"start\", to: \"llm\" },\n { from: \"llm\", to: \"tool_check\" },\n { from: \"tool_check\", to: \"bash\", label: \"yes\" },\n { from: \"tool_check\", to: \"end\", label: \"no\" },\n { from: \"bash\", to: \"append\" },\n { from: \"append\", to: \"llm\" },\n ],\n },\n s02: {\n nodes: [\n { id: \"start\", label: \"User Input\", type: \"start\", x: COL_CENTER, y: 30 },\n { id: \"llm\", label: \"LLM Call\", type: \"process\", x: COL_CENTER, y: 110 },\n { id: \"tool_check\", label: \"tool_use?\", type: \"decision\", x: COL_CENTER, y: 190 },\n { id: \"dispatch\", label: \"Tool Dispatch\", type: \"process\", x: COL_LEFT, y: 280 },\n { id: \"exec\", label: \"bash / read / write / edit\", type: \"subprocess\", x: COL_LEFT, y: 360 },\n { id: \"append\", label: \"Append Result\", type: \"process\", x: COL_LEFT, y: 440 },\n { id: \"end\", label: \"Output\", type: \"end\", x: COL_RIGHT, y: 280 },\n ],\n edges: [\n { from: \"start\", to: \"llm\" },\n { from: \"llm\", to: \"tool_check\" },\n { from: \"tool_check\", to: \"dispatch\", label: \"yes\" },\n { from: \"tool_check\", to: \"end\", label: \"no\" },\n { from: \"dispatch\", to: \"exec\" },\n { from: \"exec\", to: \"append\" },\n { from: \"append\", to: \"llm\" },\n ],\n },\n s03: {\n nodes: [\n { id: \"start\", label: \"User Input\", type: \"start\", x: COL_CENTER, y: 30 },\n { id: \"todo\", label: \"Create Todos\", type: \"process\", x: COL_CENTER, y: 100 },\n { id: \"llm\", label: \"LLM Call\", type: \"process\", x: COL_CENTER, y: 180 },\n { id: \"tool_check\", label: \"tool_use?\", type: \"decision\", x: COL_CENTER, y: 260 },\n { id: \"exec\", label: \"Execute Tool\", type: \"subprocess\", x: COL_LEFT, y: 340 },\n { id: \"append\", label: \"Append Result\", type: \"process\", x: COL_LEFT, y: 410 },\n { id: \"end\", label: \"Output\", type: \"end\", x: COL_RIGHT, y: 340 },\n ],\n edges: [\n { from: \"start\", to: \"todo\" },\n { from: \"todo\", to: \"llm\" },\n { from: \"llm\", to: \"tool_check\" },\n { from: \"tool_check\", to: \"exec\", label: \"yes\" },\n { from: \"tool_check\", to: \"end\", label: \"no\" },\n { from: \"exec\", to: \"append\" },\n { from: \"append\", to: \"llm\" },\n ],\n },\n s04: {\n nodes: [\n { id: \"start\", label: \"User Input\", type: \"start\", x: COL_CENTER, y: 30 },\n { id: \"llm\", label: \"LLM Call\", type: \"process\", x: COL_CENTER, y: 110 },\n { id: \"tool_check\", label: \"tool_use?\", type: \"decision\", x: COL_CENTER, y: 190 },\n { id: \"is_task\", label: \"task tool?\", type: \"decision\", x: COL_LEFT, y: 280 },\n { id: \"spawn\", label: \"Spawn Subagent\\n(fresh messages[])\", type: \"subprocess\", x: 60, y: 380 },\n { id: \"sub_loop\", label: \"Subagent Loop\", type: \"process\", x: 60, y: 460 },\n { id: \"exec\", label: \"Execute Tool\", type: \"subprocess\", x: COL_LEFT + 80, y: 380 },\n { id: \"append\", label: \"Append Result\", type: \"process\", x: COL_CENTER, y: 540 },\n { id: \"end\", label: \"Output\", type: \"end\", x: COL_RIGHT, y: 280 },\n ],\n edges: [\n { from: \"start\", to: \"llm\" },\n { from: \"llm\", to: \"tool_check\" },\n { from: \"tool_check\", to: \"is_task\", label: \"yes\" },\n { from: \"tool_check\", to: \"end\", label: \"no\" },\n { from: \"is_task\", to: \"spawn\", label: \"task\" },\n { from: \"is_task\", to: \"exec\", label: \"other\" },\n { from: \"spawn\", to: \"sub_loop\" },\n { from: \"sub_loop\", to: \"append\" },\n { from: \"exec\", to: \"append\" },\n { from: \"append\", to: \"llm\" },\n ],\n },\n s05: {\n nodes: [\n { id: \"start\", label: \"User Input\", type: \"start\", x: COL_CENTER, y: 30 },\n { id: \"llm\", label: \"LLM Call\", type: \"process\", x: COL_CENTER, y: 110 },\n { id: \"tool_check\", label: \"tool_use?\", type: \"decision\", x: COL_CENTER, y: 190 },\n { id: \"is_skill\", label: \"load_skill?\", type: \"decision\", x: COL_LEFT, y: 280 },\n { id: \"load\", label: \"Read SKILL.md\", type: \"subprocess\", x: 60, y: 370 },\n { id: \"inject\", label: \"Inject via\\ntool_result\", type: \"process\", x: 60, y: 450 },\n { id: \"exec\", label: \"Execute Tool\", type: \"subprocess\", x: COL_LEFT + 80, y: 370 },\n { id: \"append\", label: \"Append Result\", type: \"process\", x: COL_CENTER, y: 530 },\n { id: \"end\", label: \"Output\", type: \"end\", x: COL_RIGHT, y: 280 },\n ],\n edges: [\n { from: \"start\", to: \"llm\" },\n { from: \"llm\", to: \"tool_check\" },\n { from: \"tool_check\", to: \"is_skill\", label: \"yes\" },\n { from: \"tool_check\", to: \"end\", label: \"no\" },\n { from: \"is_skill\", to: \"load\", label: \"skill\" },\n { from: \"is_skill\", to: \"exec\", label: \"other\" },\n { from: \"load\", to: \"inject\" },\n { from: \"inject\", to: \"append\" },\n { from: \"exec\", to: \"append\" },\n { from: \"append\", to: \"llm\" },\n ],\n },\n s06: {\n nodes: [\n { id: \"start\", label: \"User Input\", type: \"start\", x: COL_CENTER, y: 30 },\n { id: \"compress_check\", label: \"Over token\\nlimit?\", type: \"decision\", x: COL_CENTER, y: 110 },\n { id: \"compress\", label: \"Compress Context\", type: \"subprocess\", x: COL_RIGHT, y: 110 },\n { id: \"llm\", label: \"LLM Call\", type: \"process\", x: COL_CENTER, y: 200 },\n { id: \"tool_check\", label: \"tool_use?\", type: \"decision\", x: COL_CENTER, y: 280 },\n { id: \"exec\", label: \"Execute Tool\", type: \"subprocess\", x: COL_LEFT, y: 360 },\n { id: \"append\", label: \"Append Result\", type: \"process\", x: COL_LEFT, y: 430 },\n { id: \"end\", label: \"Output\", type: \"end\", x: COL_RIGHT, y: 360 },\n ],\n edges: [\n { from: \"start\", to: \"compress_check\" },\n { from: \"compress_check\", to: \"compress\", label: \"yes\" },\n { from: \"compress_check\", to: \"llm\", label: \"no\" },\n { from: \"compress\", to: \"llm\" },\n { from: \"llm\", to: \"tool_check\" },\n { from: \"tool_check\", to: \"exec\", label: \"yes\" },\n { from: \"tool_check\", to: \"end\", label: \"no\" },\n { from: \"exec\", to: \"append\" },\n { from: \"append\", to: \"compress_check\" },\n ],\n },\n s07: {\n nodes: [\n { id: \"start\", label: \"User Input\", type: \"start\", x: COL_CENTER, y: 30 },\n { id: \"llm\", label: \"LLM Call\", type: \"process\", x: COL_CENTER, y: 110 },\n { id: \"tool_check\", label: \"tool_use?\", type: \"decision\", x: COL_CENTER, y: 190 },\n { id: \"is_task\", label: \"task_manager?\", type: \"decision\", x: COL_LEFT, y: 280 },\n { id: \"crud\", label: \"CRUD Task\\n(file-based)\", type: \"subprocess\", x: 60, y: 370 },\n { id: \"dep_check\", label: \"Check\\nDependencies\", type: \"process\", x: 60, y: 450 },\n { id: \"exec\", label: \"Execute Tool\", type: \"subprocess\", x: COL_LEFT + 80, y: 370 },\n { id: \"append\", label: \"Append Result\", type: \"process\", x: COL_CENTER, y: 530 },\n { id: \"end\", label: \"Output\", type: \"end\", x: COL_RIGHT, y: 280 },\n ],\n edges: [\n { from: \"start\", to: \"llm\" },\n { from: \"llm\", to: \"tool_check\" },\n { from: \"tool_check\", to: \"is_task\", label: \"yes\" },\n { from: \"tool_check\", to: \"end\", label: \"no\" },\n { from: \"is_task\", to: \"crud\", label: \"task\" },\n { from: \"is_task\", to: \"exec\", label: \"other\" },\n { from: \"crud\", to: \"dep_check\" },\n { from: \"dep_check\", to: \"append\" },\n { from: \"exec\", to: \"append\" },\n { from: \"append\", to: \"llm\" },\n ],\n },\n s08: {\n nodes: [\n { id: \"start\", label: \"User Input\", type: \"start\", x: COL_CENTER, y: 30 },\n { id: \"llm\", label: \"LLM Call\", type: \"process\", x: COL_CENTER, y: 110 },\n { id: \"tool_check\", label: \"tool_use?\", type: \"decision\", x: COL_CENTER, y: 190 },\n { id: \"bg_check\", label: \"Background?\", type: \"decision\", x: COL_LEFT, y: 280 },\n { id: \"bg_spawn\", label: \"Spawn Thread\", type: \"subprocess\", x: 60, y: 370 },\n { id: \"exec\", label: \"Execute Tool\", type: \"subprocess\", x: COL_LEFT + 80, y: 370 },\n { id: \"append\", label: \"Append Result\", type: \"process\", x: COL_CENTER, y: 450 },\n { id: \"notify\", label: \"Notification\\nQueue\", type: \"process\", x: 60, y: 450 },\n { id: \"end\", label: \"Output\", type: \"end\", x: COL_RIGHT, y: 280 },\n ],\n edges: [\n { from: \"start\", to: \"llm\" },\n { from: \"llm\", to: \"tool_check\" },\n { from: \"tool_check\", to: \"bg_check\", label: \"yes\" },\n { from: \"tool_check\", to: \"end\", label: \"no\" },\n { from: \"bg_check\", to: \"bg_spawn\", label: \"bg\" },\n { from: \"bg_check\", to: \"exec\", label: \"fg\" },\n { from: \"bg_spawn\", to: \"notify\" },\n { from: \"exec\", to: \"append\" },\n { from: \"append\", to: \"llm\" },\n { from: \"notify\", to: \"llm\" },\n ],\n },\n s09: {\n nodes: [\n { id: \"start\", label: \"User Input\", type: \"start\", x: COL_CENTER, y: 30 },\n { id: \"llm\", label: \"LLM Call\\n(team lead)\", type: \"process\", x: COL_CENTER, y: 110 },\n { id: \"tool_check\", label: \"tool_use?\", type: \"decision\", x: COL_CENTER, y: 200 },\n { id: \"is_team\", label: \"Team tool?\", type: \"decision\", x: COL_LEFT, y: 290 },\n { id: \"spawn\", label: \"Spawn\\nTeammate\", type: \"subprocess\", x: 60, y: 390 },\n { id: \"msg\", label: \"Send Message\\n(JSONL inbox)\", type: \"subprocess\", x: 60, y: 470 },\n { id: \"exec\", label: \"Execute Tool\", type: \"subprocess\", x: COL_LEFT + 80, y: 390 },\n { id: \"append\", label: \"Append Result\", type: \"process\", x: COL_CENTER, y: 550 },\n { id: \"end\", label: \"Output\", type: \"end\", x: COL_RIGHT, y: 290 },\n { id: \"teammate\", label: \"Teammate Agent\\n(own loop)\", type: \"process\", x: COL_RIGHT, y: 470 },\n ],\n edges: [\n { from: \"start\", to: \"llm\" },\n { from: \"llm\", to: \"tool_check\" },\n { from: \"tool_check\", to: \"is_team\", label: \"yes\" },\n { from: \"tool_check\", to: \"end\", label: \"no\" },\n { from: \"is_team\", to: \"spawn\", label: \"spawn\" },\n { from: \"is_team\", to: \"exec\", label: \"other\" },\n { from: \"spawn\", to: \"teammate\" },\n { from: \"spawn\", to: \"msg\" },\n { from: \"msg\", to: \"append\" },\n { from: \"exec\", to: \"append\" },\n { from: \"append\", to: \"llm\" },\n ],\n },\n s10: {\n nodes: [\n { id: \"start\", label: \"User Input\", type: \"start\", x: COL_CENTER, y: 30 },\n { id: \"llm\", label: \"LLM Call\\n(team lead)\", type: \"process\", x: COL_CENTER, y: 110 },\n { id: \"tool_check\", label: \"tool_use?\", type: \"decision\", x: COL_CENTER, y: 200 },\n { id: \"is_proto\", label: \"Protocol?\", type: \"decision\", x: COL_LEFT, y: 290 },\n { id: \"shutdown\", label: \"Shutdown\\nRequest\", type: \"subprocess\", x: 60, y: 390 },\n { id: \"fsm\", label: \"FSM:\\npending->approved\", type: \"process\", x: 60, y: 470 },\n { id: \"exec\", label: \"Execute Tool\", type: \"subprocess\", x: COL_LEFT + 80, y: 390 },\n { id: \"append\", label: \"Append Result\", type: \"process\", x: COL_CENTER, y: 550 },\n { id: \"end\", label: \"Output\", type: \"end\", x: COL_RIGHT, y: 290 },\n { id: \"teammate\", label: \"Teammate\\nreceives request_id\", type: \"process\", x: COL_RIGHT, y: 470 },\n ],\n edges: [\n { from: \"start\", to: \"llm\" },\n { from: \"llm\", to: \"tool_check\" },\n { from: \"tool_check\", to: \"is_proto\", label: \"yes\" },\n { from: \"tool_check\", to: \"end\", label: \"no\" },\n { from: \"is_proto\", to: \"shutdown\", label: \"shutdown\" },\n { from: \"is_proto\", to: \"exec\", label: \"other\" },\n { from: \"shutdown\", to: \"fsm\" },\n { from: \"fsm\", to: \"teammate\" },\n { from: \"teammate\", to: \"append\" },\n { from: \"exec\", to: \"append\" },\n { from: \"append\", to: \"llm\" },\n ],\n },\n s11: {\n nodes: [\n { id: \"start\", label: \"User Input\", type: \"start\", x: COL_CENTER, y: 30 },\n { id: \"inbox\", label: \"Check Inbox\", type: \"process\", x: COL_CENTER, y: 100 },\n { id: \"llm\", label: \"LLM Call\", type: \"process\", x: COL_CENTER, y: 180 },\n { id: \"tool_check\", label: \"tool_use?\", type: \"decision\", x: COL_CENTER, y: 260 },\n { id: \"exec\", label: \"Execute Tool\", type: \"subprocess\", x: COL_LEFT, y: 340 },\n { id: \"append\", label: \"Append Result\", type: \"process\", x: COL_LEFT, y: 410 },\n { id: \"end\", label: \"Output\", type: \"end\", x: COL_RIGHT, y: 340 },\n { id: \"idle\", label: \"Idle Cycle\", type: \"process\", x: COL_RIGHT, y: 420 },\n { id: \"poll\", label: \"Poll Tasks\\n+ Auto-Claim\", type: \"subprocess\", x: COL_RIGHT, y: 500 },\n ],\n edges: [\n { from: \"start\", to: \"inbox\" },\n { from: \"inbox\", to: \"llm\" },\n { from: \"llm\", to: \"tool_check\" },\n { from: \"tool_check\", to: \"exec\", label: \"yes\" },\n { from: \"tool_check\", to: \"end\", label: \"no\" },\n { from: \"exec\", to: \"append\" },\n { from: \"append\", to: \"llm\" },\n { from: \"end\", to: \"idle\" },\n { from: \"idle\", to: \"poll\" },\n { from: \"poll\", to: \"inbox\" },\n ],\n },\n s12: {\n nodes: [\n { id: \"start\", label: \"User Input\", type: \"start\", x: COL_CENTER, y: 30 },\n { id: \"llm\", label: \"LLM Call\", type: \"process\", x: COL_CENTER, y: 110 },\n { id: \"tool_check\", label: \"tool_use?\", type: \"decision\", x: COL_CENTER, y: 190 },\n { id: \"is_wt\", label: \"worktree tool?\", type: \"decision\", x: COL_LEFT, y: 280 },\n { id: \"task\", label: \"Task Board\\\\n(.tasks)\", type: \"process\", x: 60, y: 360 },\n { id: \"wt_create\", label: \"Allocate / Enter\\\\nWorktree\", type: \"subprocess\", x: 60, y: 440 },\n { id: \"wt_run\", label: \"Run in\\\\nIsolated Dir\", type: \"subprocess\", x: COL_LEFT + 80, y: 360 },\n { id: \"wt_close\", label: \"Closeout:\\\\nworktree_keep / remove\", type: \"process\", x: COL_LEFT + 80, y: 440 },\n { id: \"events\", label: \"Emit Lifecycle Events\\\\n(side-channel)\", type: \"process\", x: COL_RIGHT, y: 420 },\n { id: \"events_read\", label: \"Optional Read\\\\nworktree_events\", type: \"subprocess\", x: COL_RIGHT, y: 520 },\n { id: \"append\", label: \"Append Result\", type: \"process\", x: COL_CENTER, y: 530 },\n { id: \"end\", label: \"Output\", type: \"end\", x: COL_RIGHT, y: 280 },\n ],\n edges: [\n { from: \"start\", to: \"llm\" },\n { from: \"llm\", to: \"tool_check\" },\n { from: \"tool_check\", to: \"is_wt\", label: \"yes\" },\n { from: \"tool_check\", to: \"end\", label: \"no\" },\n { from: \"is_wt\", to: \"task\", label: \"task ops\" },\n { from: \"is_wt\", to: \"wt_create\", label: \"create/bind\" },\n { from: \"is_wt\", to: \"wt_run\", label: \"run/status\" },\n { from: \"task\", to: \"wt_create\", label: \"allocate lane\" },\n { from: \"wt_create\", to: \"wt_run\" },\n { from: \"task\", to: \"append\", label: \"task result\" },\n { from: \"wt_create\", to: \"events\", label: \"emit create\" },\n { from: \"wt_create\", to: \"append\", label: \"create result\" },\n { from: \"wt_run\", to: \"wt_close\" },\n { from: \"wt_run\", to: \"append\", label: \"run/status result\" },\n { from: \"wt_close\", to: \"events\", label: \"emit closeout\" },\n { from: \"wt_close\", to: \"append\", label: \"closeout result\" },\n { from: \"events\", to: \"events_read\", label: \"optional query\" },\n { from: \"events_read\", to: \"append\", label: \"events result\" },\n { from: \"append\", to: \"llm\" },\n ],\n },\n};\n\nexport function getFlowForVersion(version: string): FlowDefinition | null {\n return EXECUTION_FLOWS[version] ?? null;\n}\n"
},
{
"path": "web/src/data/generated/docs.json",
"content": "[\n {\n \"version\": \"s01\",\n \"locale\": \"en\",\n \"title\": \"s01: The Agent Loop\",\n \"content\": \"# s01: The Agent Loop\\n\\n`[ s01 ] s02 > s03 > s04 > s05 > s06 | s07 > s08 > s09 > s10 > s11 > s12`\\n\\n> *\\\"One loop & Bash is all you need\\\"* -- one tool + one loop = an agent.\\n\\n## Problem\\n\\nA language model can reason about code, but it can't *touch* the real world -- can't read files, run tests, or check errors. Without a loop, every tool call requires you to manually copy-paste results back. You become the loop.\\n\\n## Solution\\n\\n```\\n+--------+ +-------+ +---------+\\n| User | ---> | LLM | ---> | Tool |\\n| prompt | | | | execute |\\n+--------+ +---+---+ +----+----+\\n ^ |\\n | tool_result |\\n +----------------+\\n (loop until stop_reason != \\\"tool_use\\\")\\n```\\n\\nOne exit condition controls the entire flow. The loop runs until the model stops calling tools.\\n\\n## How It Works\\n\\n1. User prompt becomes the first message.\\n\\n```python\\nmessages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": query})\\n```\\n\\n2. Send messages + tool definitions to the LLM.\\n\\n```python\\nresponse = client.messages.create(\\n model=MODEL, system=SYSTEM, messages=messages,\\n tools=TOOLS, max_tokens=8000,\\n)\\n```\\n\\n3. Append the assistant response. Check `stop_reason` -- if the model didn't call a tool, we're done.\\n\\n```python\\nmessages.append({\\\"role\\\": \\\"assistant\\\", \\\"content\\\": response.content})\\nif response.stop_reason != \\\"tool_use\\\":\\n return\\n```\\n\\n4. Execute each tool call, collect results, append as a user message. Loop back to step 2.\\n\\n```python\\nresults = []\\nfor block in response.content:\\n if block.type == \\\"tool_use\\\":\\n output = run_bash(block.input[\\\"command\\\"])\\n results.append({\\n \\\"type\\\": \\\"tool_result\\\",\\n \\\"tool_use_id\\\": block.id,\\n \\\"content\\\": output,\\n })\\nmessages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": results})\\n```\\n\\nAssembled into one function:\\n\\n```python\\ndef agent_loop(query):\\n messages = [{\\\"role\\\": \\\"user\\\", \\\"content\\\": query}]\\n while True:\\n response = client.messages.create(\\n model=MODEL, system=SYSTEM, messages=messages,\\n tools=TOOLS, max_tokens=8000,\\n )\\n messages.append({\\\"role\\\": \\\"assistant\\\", \\\"content\\\": response.content})\\n\\n if response.stop_reason != \\\"tool_use\\\":\\n return\\n\\n results = []\\n for block in response.content:\\n if block.type == \\\"tool_use\\\":\\n output = run_bash(block.input[\\\"command\\\"])\\n results.append({\\n \\\"type\\\": \\\"tool_result\\\",\\n \\\"tool_use_id\\\": block.id,\\n \\\"content\\\": output,\\n })\\n messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": results})\\n```\\n\\nThat's the entire agent in under 30 lines. Everything else in this course layers on top -- without changing the loop.\\n\\n## What Changed\\n\\n| Component | Before | After |\\n|---------------|------------|--------------------------------|\\n| Agent loop | (none) | `while True` + stop_reason |\\n| Tools | (none) | `bash` (one tool) |\\n| Messages | (none) | Accumulating list |\\n| Control flow | (none) | `stop_reason != \\\"tool_use\\\"` |\\n\\n## Try It\\n\\n```sh\\ncd learn-claude-code\\npython agents/s01_agent_loop.py\\n```\\n\\n1. `Create a file called hello.py that prints \\\"Hello, World!\\\"`\\n2. `List all Python files in this directory`\\n3. `What is the current git branch?`\\n4. `Create a directory called test_output and write 3 files in it`\\n\"\n },\n {\n \"version\": \"s02\",\n \"locale\": \"en\",\n \"title\": \"s02: Tool Use\",\n \"content\": \"# s02: Tool Use\\n\\n`s01 > [ s02 ] s03 > s04 > s05 > s06 | s07 > s08 > s09 > s10 > s11 > s12`\\n\\n> *\\\"Adding a tool means adding one handler\\\"* -- the loop stays the same; new tools register into the dispatch map.\\n\\n## Problem\\n\\nWith only `bash`, the agent shells out for everything. `cat` truncates unpredictably, `sed` fails on special characters, and every bash call is an unconstrained security surface. Dedicated tools like `read_file` and `write_file` let you enforce path sandboxing at the tool level.\\n\\nThe key insight: adding tools does not require changing the loop.\\n\\n## Solution\\n\\n```\\n+--------+ +-------+ +------------------+\\n| User | ---> | LLM | ---> | Tool Dispatch |\\n| prompt | | | | { |\\n+--------+ +---+---+ | bash: run_bash |\\n ^ | read: run_read |\\n | | write: run_wr |\\n +-----------+ edit: run_edit |\\n tool_result | } |\\n +------------------+\\n\\nThe dispatch map is a dict: {tool_name: handler_function}.\\nOne lookup replaces any if/elif chain.\\n```\\n\\n## How It Works\\n\\n1. Each tool gets a handler function. Path sandboxing prevents workspace escape.\\n\\n```python\\ndef safe_path(p: str) -> Path:\\n path = (WORKDIR / p).resolve()\\n if not path.is_relative_to(WORKDIR):\\n raise ValueError(f\\\"Path escapes workspace: {p}\\\")\\n return path\\n\\ndef run_read(path: str, limit: int = None) -> str:\\n text = safe_path(path).read_text()\\n lines = text.splitlines()\\n if limit and limit < len(lines):\\n lines = lines[:limit]\\n return \\\"\\\\n\\\".join(lines)[:50000]\\n```\\n\\n2. The dispatch map links tool names to handlers.\\n\\n```python\\nTOOL_HANDLERS = {\\n \\\"bash\\\": lambda **kw: run_bash(kw[\\\"command\\\"]),\\n \\\"read_file\\\": lambda **kw: run_read(kw[\\\"path\\\"], kw.get(\\\"limit\\\")),\\n \\\"write_file\\\": lambda **kw: run_write(kw[\\\"path\\\"], kw[\\\"content\\\"]),\\n \\\"edit_file\\\": lambda **kw: run_edit(kw[\\\"path\\\"], kw[\\\"old_text\\\"],\\n kw[\\\"new_text\\\"]),\\n}\\n```\\n\\n3. In the loop, look up the handler by name. The loop body itself is unchanged from s01.\\n\\n```python\\nfor block in response.content:\\n if block.type == \\\"tool_use\\\":\\n handler = TOOL_HANDLERS.get(block.name)\\n output = handler(**block.input) if handler \\\\\\n else f\\\"Unknown tool: {block.name}\\\"\\n results.append({\\n \\\"type\\\": \\\"tool_result\\\",\\n \\\"tool_use_id\\\": block.id,\\n \\\"content\\\": output,\\n })\\n```\\n\\nAdd a tool = add a handler + add a schema entry. The loop never changes.\\n\\n## What Changed From s01\\n\\n| Component | Before (s01) | After (s02) |\\n|----------------|--------------------|----------------------------|\\n| Tools | 1 (bash only) | 4 (bash, read, write, edit)|\\n| Dispatch | Hardcoded bash call | `TOOL_HANDLERS` dict |\\n| Path safety | None | `safe_path()` sandbox |\\n| Agent loop | Unchanged | Unchanged |\\n\\n## Try It\\n\\n```sh\\ncd learn-claude-code\\npython agents/s02_tool_use.py\\n```\\n\\n1. `Read the file requirements.txt`\\n2. `Create a file called greet.py with a greet(name) function`\\n3. `Edit greet.py to add a docstring to the function`\\n4. `Read greet.py to verify the edit worked`\\n\"\n },\n {\n \"version\": \"s03\",\n \"locale\": \"en\",\n \"title\": \"s03: TodoWrite\",\n \"content\": \"# s03: TodoWrite\\n\\n`s01 > s02 > [ s03 ] s04 > s05 > s06 | s07 > s08 > s09 > s10 > s11 > s12`\\n\\n> *\\\"An agent without a plan drifts\\\"* -- list the steps first, then execute.\\n\\n## Problem\\n\\nOn multi-step tasks, the model loses track. It repeats work, skips steps, or wanders off. Long conversations make this worse -- the system prompt fades as tool results fill the context. A 10-step refactoring might complete steps 1-3, then the model starts improvising because it forgot steps 4-10.\\n\\n## Solution\\n\\n```\\n+--------+ +-------+ +---------+\\n| User | ---> | LLM | ---> | Tools |\\n| prompt | | | | + todo |\\n+--------+ +---+---+ +----+----+\\n ^ |\\n | tool_result |\\n +----------------+\\n |\\n +-----------+-----------+\\n | TodoManager state |\\n | [ ] task A |\\n | [>] task B <- doing |\\n | [x] task C |\\n +-----------------------+\\n |\\n if rounds_since_todo >= 3:\\n inject into tool_result\\n```\\n\\n## How It Works\\n\\n1. TodoManager stores items with statuses. Only one item can be `in_progress` at a time.\\n\\n```python\\nclass TodoManager:\\n def update(self, items: list) -> str:\\n validated, in_progress_count = [], 0\\n for item in items:\\n status = item.get(\\\"status\\\", \\\"pending\\\")\\n if status == \\\"in_progress\\\":\\n in_progress_count += 1\\n validated.append({\\\"id\\\": item[\\\"id\\\"], \\\"text\\\": item[\\\"text\\\"],\\n \\\"status\\\": status})\\n if in_progress_count > 1:\\n raise ValueError(\\\"Only one task can be in_progress\\\")\\n self.items = validated\\n return self.render()\\n```\\n\\n2. The `todo` tool goes into the dispatch map like any other tool.\\n\\n```python\\nTOOL_HANDLERS = {\\n # ...base tools...\\n \\\"todo\\\": lambda **kw: TODO.update(kw[\\\"items\\\"]),\\n}\\n```\\n\\n3. A nag reminder injects a nudge if the model goes 3+ rounds without calling `todo`.\\n\\n```python\\nif rounds_since_todo >= 3 and messages:\\n last = messages[-1]\\n if last[\\\"role\\\"] == \\\"user\\\" and isinstance(last.get(\\\"content\\\"), list):\\n last[\\\"content\\\"].insert(0, {\\n \\\"type\\\": \\\"text\\\",\\n \\\"text\\\": \\\"Update your todos. \\\",\\n })\\n```\\n\\nThe \\\"one in_progress at a time\\\" constraint forces sequential focus. The nag reminder creates accountability.\\n\\n## What Changed From s02\\n\\n| Component | Before (s02) | After (s03) |\\n|----------------|------------------|----------------------------|\\n| Tools | 4 | 5 (+todo) |\\n| Planning | None | TodoManager with statuses |\\n| Nag injection | None | `` after 3 rounds|\\n| Agent loop | Simple dispatch | + rounds_since_todo counter|\\n\\n## Try It\\n\\n```sh\\ncd learn-claude-code\\npython agents/s03_todo_write.py\\n```\\n\\n1. `Refactor the file hello.py: add type hints, docstrings, and a main guard`\\n2. `Create a Python package with __init__.py, utils.py, and tests/test_utils.py`\\n3. `Review all Python files and fix any style issues`\\n\"\n },\n {\n \"version\": \"s04\",\n \"locale\": \"en\",\n \"title\": \"s04: Subagents\",\n \"content\": \"# s04: Subagents\\n\\n`s01 > s02 > s03 > [ s04 ] s05 > s06 | s07 > s08 > s09 > s10 > s11 > s12`\\n\\n> *\\\"Break big tasks down; each subtask gets a clean context\\\"* -- subagents use independent messages[], keeping the main conversation clean.\\n\\n## Problem\\n\\nAs the agent works, its messages array grows. Every file read, every bash output stays in context permanently. \\\"What testing framework does this project use?\\\" might require reading 5 files, but the parent only needs the answer: \\\"pytest.\\\"\\n\\n## Solution\\n\\n```\\nParent agent Subagent\\n+------------------+ +------------------+\\n| messages=[...] | | messages=[] | <-- fresh\\n| | dispatch | |\\n| tool: task | ----------> | while tool_use: |\\n| prompt=\\\"...\\\" | | call tools |\\n| | summary | append results |\\n| result = \\\"...\\\" | <---------- | return last text |\\n+------------------+ +------------------+\\n\\nParent context stays clean. Subagent context is discarded.\\n```\\n\\n## How It Works\\n\\n1. The parent gets a `task` tool. The child gets all base tools except `task` (no recursive spawning).\\n\\n```python\\nPARENT_TOOLS = CHILD_TOOLS + [\\n {\\\"name\\\": \\\"task\\\",\\n \\\"description\\\": \\\"Spawn a subagent with fresh context.\\\",\\n \\\"input_schema\\\": {\\n \\\"type\\\": \\\"object\\\",\\n \\\"properties\\\": {\\\"prompt\\\": {\\\"type\\\": \\\"string\\\"}},\\n \\\"required\\\": [\\\"prompt\\\"],\\n }},\\n]\\n```\\n\\n2. The subagent starts with `messages=[]` and runs its own loop. Only the final text returns to the parent.\\n\\n```python\\ndef run_subagent(prompt: str) -> str:\\n sub_messages = [{\\\"role\\\": \\\"user\\\", \\\"content\\\": prompt}]\\n for _ in range(30): # safety limit\\n response = client.messages.create(\\n model=MODEL, system=SUBAGENT_SYSTEM,\\n messages=sub_messages,\\n tools=CHILD_TOOLS, max_tokens=8000,\\n )\\n sub_messages.append({\\\"role\\\": \\\"assistant\\\",\\n \\\"content\\\": response.content})\\n if response.stop_reason != \\\"tool_use\\\":\\n break\\n results = []\\n for block in response.content:\\n if block.type == \\\"tool_use\\\":\\n handler = TOOL_HANDLERS.get(block.name)\\n output = handler(**block.input)\\n results.append({\\\"type\\\": \\\"tool_result\\\",\\n \\\"tool_use_id\\\": block.id,\\n \\\"content\\\": str(output)[:50000]})\\n sub_messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": results})\\n return \\\"\\\".join(\\n b.text for b in response.content if hasattr(b, \\\"text\\\")\\n ) or \\\"(no summary)\\\"\\n```\\n\\nThe child's entire message history (possibly 30+ tool calls) is discarded. The parent receives a one-paragraph summary as a normal `tool_result`.\\n\\n## What Changed From s03\\n\\n| Component | Before (s03) | After (s04) |\\n|----------------|------------------|---------------------------|\\n| Tools | 5 | 5 (base) + task (parent) |\\n| Context | Single shared | Parent + child isolation |\\n| Subagent | None | `run_subagent()` function |\\n| Return value | N/A | Summary text only |\\n\\n## Try It\\n\\n```sh\\ncd learn-claude-code\\npython agents/s04_subagent.py\\n```\\n\\n1. `Use a subtask to find what testing framework this project uses`\\n2. `Delegate: read all .py files and summarize what each one does`\\n3. `Use a task to create a new module, then verify it from here`\\n\"\n },\n {\n \"version\": \"s05\",\n \"locale\": \"en\",\n \"title\": \"s05: Skills\",\n \"content\": \"# s05: Skills\\n\\n`s01 > s02 > s03 > s04 > [ s05 ] s06 | s07 > s08 > s09 > s10 > s11 > s12`\\n\\n> *\\\"Load knowledge when you need it, not upfront\\\"* -- inject via tool_result, not the system prompt.\\n\\n## Problem\\n\\nYou want the agent to follow domain-specific workflows: git conventions, testing patterns, code review checklists. Putting everything in the system prompt wastes tokens on unused skills. 10 skills at 2000 tokens each = 20,000 tokens, most of which are irrelevant to any given task.\\n\\n## Solution\\n\\n```\\nSystem prompt (Layer 1 -- always present):\\n+--------------------------------------+\\n| You are a coding agent. |\\n| Skills available: |\\n| - git: Git workflow helpers | ~100 tokens/skill\\n| - test: Testing best practices |\\n+--------------------------------------+\\n\\nWhen model calls load_skill(\\\"git\\\"):\\n+--------------------------------------+\\n| tool_result (Layer 2 -- on demand): |\\n| |\\n| Full git workflow instructions... | ~2000 tokens\\n| Step 1: ... |\\n| |\\n+--------------------------------------+\\n```\\n\\nLayer 1: skill *names* in system prompt (cheap). Layer 2: full *body* via tool_result (on demand).\\n\\n## How It Works\\n\\n1. Each skill is a directory containing a `SKILL.md` with YAML frontmatter.\\n\\n```\\nskills/\\n pdf/\\n SKILL.md # ---\\\\n name: pdf\\\\n description: Process PDF files\\\\n ---\\\\n ...\\n code-review/\\n SKILL.md # ---\\\\n name: code-review\\\\n description: Review code\\\\n ---\\\\n ...\\n```\\n\\n2. SkillLoader scans for `SKILL.md` files, uses the directory name as the skill identifier.\\n\\n```python\\nclass SkillLoader:\\n def __init__(self, skills_dir: Path):\\n self.skills = {}\\n for f in sorted(skills_dir.rglob(\\\"SKILL.md\\\")):\\n text = f.read_text()\\n meta, body = self._parse_frontmatter(text)\\n name = meta.get(\\\"name\\\", f.parent.name)\\n self.skills[name] = {\\\"meta\\\": meta, \\\"body\\\": body}\\n\\n def get_descriptions(self) -> str:\\n lines = []\\n for name, skill in self.skills.items():\\n desc = skill[\\\"meta\\\"].get(\\\"description\\\", \\\"\\\")\\n lines.append(f\\\" - {name}: {desc}\\\")\\n return \\\"\\\\n\\\".join(lines)\\n\\n def get_content(self, name: str) -> str:\\n skill = self.skills.get(name)\\n if not skill:\\n return f\\\"Error: Unknown skill '{name}'.\\\"\\n return f\\\"\\\\n{skill['body']}\\\\n \\\"\\n```\\n\\n3. Layer 1 goes into the system prompt. Layer 2 is just another tool handler.\\n\\n```python\\nSYSTEM = f\\\"\\\"\\\"You are a coding agent at {WORKDIR}.\\nSkills available:\\n{SKILL_LOADER.get_descriptions()}\\\"\\\"\\\"\\n\\nTOOL_HANDLERS = {\\n # ...base tools...\\n \\\"load_skill\\\": lambda **kw: SKILL_LOADER.get_content(kw[\\\"name\\\"]),\\n}\\n```\\n\\nThe model learns what skills exist (cheap) and loads them when relevant (expensive).\\n\\n## What Changed From s04\\n\\n| Component | Before (s04) | After (s05) |\\n|----------------|------------------|----------------------------|\\n| Tools | 5 (base + task) | 5 (base + load_skill) |\\n| System prompt | Static string | + skill descriptions |\\n| Knowledge | None | skills/\\\\*/SKILL.md files |\\n| Injection | None | Two-layer (system + result)|\\n\\n## Try It\\n\\n```sh\\ncd learn-claude-code\\npython agents/s05_skill_loading.py\\n```\\n\\n1. `What skills are available?`\\n2. `Load the agent-builder skill and follow its instructions`\\n3. `I need to do a code review -- load the relevant skill first`\\n4. `Build an MCP server using the mcp-builder skill`\\n\"\n },\n {\n \"version\": \"s06\",\n \"locale\": \"en\",\n \"title\": \"s06: Context Compact\",\n \"content\": \"# s06: Context Compact\\n\\n`s01 > s02 > s03 > s04 > s05 > [ s06 ] | s07 > s08 > s09 > s10 > s11 > s12`\\n\\n> *\\\"Context will fill up; you need a way to make room\\\"* -- three-layer compression strategy for infinite sessions.\\n\\n## Problem\\n\\nThe context window is finite. A single `read_file` on a 1000-line file costs ~4000 tokens. After reading 30 files and running 20 bash commands, you hit 100,000+ tokens. The agent cannot work on large codebases without compression.\\n\\n## Solution\\n\\nThree layers, increasing in aggressiveness:\\n\\n```\\nEvery turn:\\n+------------------+\\n| Tool call result |\\n+------------------+\\n |\\n v\\n[Layer 1: micro_compact] (silent, every turn)\\n Replace tool_result > 3 turns old\\n with \\\"[Previous: used {tool_name}]\\\"\\n |\\n v\\n[Check: tokens > 50000?]\\n | |\\n no yes\\n | |\\n v v\\ncontinue [Layer 2: auto_compact]\\n Save transcript to .transcripts/\\n LLM summarizes conversation.\\n Replace all messages with [summary].\\n |\\n v\\n [Layer 3: compact tool]\\n Model calls compact explicitly.\\n Same summarization as auto_compact.\\n```\\n\\n## How It Works\\n\\n1. **Layer 1 -- micro_compact**: Before each LLM call, replace old tool results with placeholders.\\n\\n```python\\ndef micro_compact(messages: list) -> list:\\n tool_results = []\\n for i, msg in enumerate(messages):\\n if msg[\\\"role\\\"] == \\\"user\\\" and isinstance(msg.get(\\\"content\\\"), list):\\n for j, part in enumerate(msg[\\\"content\\\"]):\\n if isinstance(part, dict) and part.get(\\\"type\\\") == \\\"tool_result\\\":\\n tool_results.append((i, j, part))\\n if len(tool_results) <= KEEP_RECENT:\\n return messages\\n for _, _, part in tool_results[:-KEEP_RECENT]:\\n if len(part.get(\\\"content\\\", \\\"\\\")) > 100:\\n part[\\\"content\\\"] = f\\\"[Previous: used {tool_name}]\\\"\\n return messages\\n```\\n\\n2. **Layer 2 -- auto_compact**: When tokens exceed threshold, save full transcript to disk, then ask the LLM to summarize.\\n\\n```python\\ndef auto_compact(messages: list) -> list:\\n # Save transcript for recovery\\n transcript_path = TRANSCRIPT_DIR / f\\\"transcript_{int(time.time())}.jsonl\\\"\\n with open(transcript_path, \\\"w\\\") as f:\\n for msg in messages:\\n f.write(json.dumps(msg, default=str) + \\\"\\\\n\\\")\\n # LLM summarizes\\n response = client.messages.create(\\n model=MODEL,\\n messages=[{\\\"role\\\": \\\"user\\\", \\\"content\\\":\\n \\\"Summarize this conversation for continuity...\\\"\\n + json.dumps(messages, default=str)[:80000]}],\\n max_tokens=2000,\\n )\\n return [\\n {\\\"role\\\": \\\"user\\\", \\\"content\\\": f\\\"[Compressed]\\\\n\\\\n{response.content[0].text}\\\"},\\n {\\\"role\\\": \\\"assistant\\\", \\\"content\\\": \\\"Understood. Continuing.\\\"},\\n ]\\n```\\n\\n3. **Layer 3 -- manual compact**: The `compact` tool triggers the same summarization on demand.\\n\\n4. The loop integrates all three:\\n\\n```python\\ndef agent_loop(messages: list):\\n while True:\\n micro_compact(messages) # Layer 1\\n if estimate_tokens(messages) > THRESHOLD:\\n messages[:] = auto_compact(messages) # Layer 2\\n response = client.messages.create(...)\\n # ... tool execution ...\\n if manual_compact:\\n messages[:] = auto_compact(messages) # Layer 3\\n```\\n\\nTranscripts preserve full history on disk. Nothing is truly lost -- just moved out of active context.\\n\\n## What Changed From s05\\n\\n| Component | Before (s05) | After (s06) |\\n|----------------|------------------|----------------------------|\\n| Tools | 5 | 5 (base + compact) |\\n| Context mgmt | None | Three-layer compression |\\n| Micro-compact | None | Old results -> placeholders|\\n| Auto-compact | None | Token threshold trigger |\\n| Transcripts | None | Saved to .transcripts/ |\\n\\n## Try It\\n\\n```sh\\ncd learn-claude-code\\npython agents/s06_context_compact.py\\n```\\n\\n1. `Read every Python file in the agents/ directory one by one` (watch micro-compact replace old results)\\n2. `Keep reading files until compression triggers automatically`\\n3. `Use the compact tool to manually compress the conversation`\\n\"\n },\n {\n \"version\": \"s07\",\n \"locale\": \"en\",\n \"title\": \"s07: Task System\",\n \"content\": \"# s07: Task System\\n\\n`s01 > s02 > s03 > s04 > s05 > s06 | [ s07 ] s08 > s09 > s10 > s11 > s12`\\n\\n> *\\\"Break big goals into small tasks, order them, persist to disk\\\"* -- a file-based task graph with dependencies, laying the foundation for multi-agent collaboration.\\n\\n## Problem\\n\\ns03's TodoManager is a flat checklist in memory: no ordering, no dependencies, no status beyond done-or-not. Real goals have structure -- task B depends on task A, tasks C and D can run in parallel, task E waits for both C and D.\\n\\nWithout explicit relationships, the agent can't tell what's ready, what's blocked, or what can run concurrently. And because the list lives only in memory, context compression (s06) wipes it clean.\\n\\n## Solution\\n\\nPromote the checklist into a **task graph** persisted to disk. Each task is a JSON file with status, dependencies (`blockedBy`), and dependents (`blocks`). The graph answers three questions at any moment:\\n\\n- **What's ready?** -- tasks with `pending` status and empty `blockedBy`.\\n- **What's blocked?** -- tasks waiting on unfinished dependencies.\\n- **What's done?** -- `completed` tasks, whose completion automatically unblocks dependents.\\n\\n```\\n.tasks/\\n task_1.json {\\\"id\\\":1, \\\"status\\\":\\\"completed\\\"}\\n task_2.json {\\\"id\\\":2, \\\"blockedBy\\\":[1], \\\"status\\\":\\\"pending\\\"}\\n task_3.json {\\\"id\\\":3, \\\"blockedBy\\\":[1], \\\"status\\\":\\\"pending\\\"}\\n task_4.json {\\\"id\\\":4, \\\"blockedBy\\\":[2,3], \\\"status\\\":\\\"pending\\\"}\\n\\nTask graph (DAG):\\n +----------+\\n +--> | task 2 | --+\\n | | pending | |\\n+----------+ +----------+ +--> +----------+\\n| task 1 | | task 4 |\\n| completed| --> +----------+ +--> | blocked |\\n+----------+ | task 3 | --+ +----------+\\n | pending |\\n +----------+\\n\\nOrdering: task 1 must finish before 2 and 3\\nParallelism: tasks 2 and 3 can run at the same time\\nDependencies: task 4 waits for both 2 and 3\\nStatus: pending -> in_progress -> completed\\n```\\n\\nThis task graph becomes the coordination backbone for everything after s07: background execution (s08), multi-agent teams (s09+), and worktree isolation (s12) all read from and write to this same structure.\\n\\n## How It Works\\n\\n1. **TaskManager**: one JSON file per task, CRUD with dependency graph.\\n\\n```python\\nclass TaskManager:\\n def __init__(self, tasks_dir: Path):\\n self.dir = tasks_dir\\n self.dir.mkdir(exist_ok=True)\\n self._next_id = self._max_id() + 1\\n\\n def create(self, subject, description=\\\"\\\"):\\n task = {\\\"id\\\": self._next_id, \\\"subject\\\": subject,\\n \\\"status\\\": \\\"pending\\\", \\\"blockedBy\\\": [],\\n \\\"blocks\\\": [], \\\"owner\\\": \\\"\\\"}\\n self._save(task)\\n self._next_id += 1\\n return json.dumps(task, indent=2)\\n```\\n\\n2. **Dependency resolution**: completing a task clears its ID from every other task's `blockedBy` list, automatically unblocking dependents.\\n\\n```python\\ndef _clear_dependency(self, completed_id):\\n for f in self.dir.glob(\\\"task_*.json\\\"):\\n task = json.loads(f.read_text())\\n if completed_id in task.get(\\\"blockedBy\\\", []):\\n task[\\\"blockedBy\\\"].remove(completed_id)\\n self._save(task)\\n```\\n\\n3. **Status + dependency wiring**: `update` handles transitions and dependency edges.\\n\\n```python\\ndef update(self, task_id, status=None,\\n add_blocked_by=None, add_blocks=None):\\n task = self._load(task_id)\\n if status:\\n task[\\\"status\\\"] = status\\n if status == \\\"completed\\\":\\n self._clear_dependency(task_id)\\n self._save(task)\\n```\\n\\n4. Four task tools go into the dispatch map.\\n\\n```python\\nTOOL_HANDLERS = {\\n # ...base tools...\\n \\\"task_create\\\": lambda **kw: TASKS.create(kw[\\\"subject\\\"]),\\n \\\"task_update\\\": lambda **kw: TASKS.update(kw[\\\"task_id\\\"], kw.get(\\\"status\\\")),\\n \\\"task_list\\\": lambda **kw: TASKS.list_all(),\\n \\\"task_get\\\": lambda **kw: TASKS.get(kw[\\\"task_id\\\"]),\\n}\\n```\\n\\nFrom s07 onward, the task graph is the default for multi-step work. s03's Todo remains for quick single-session checklists.\\n\\n## What Changed From s06\\n\\n| Component | Before (s06) | After (s07) |\\n|---|---|---|\\n| Tools | 5 | 8 (`task_create/update/list/get`) |\\n| Planning model | Flat checklist (in-memory) | Task graph with dependencies (on disk) |\\n| Relationships | None | `blockedBy` + `blocks` edges |\\n| Status tracking | Done or not | `pending` -> `in_progress` -> `completed` |\\n| Persistence | Lost on compression | Survives compression and restarts |\\n\\n## Try It\\n\\n```sh\\ncd learn-claude-code\\npython agents/s07_task_system.py\\n```\\n\\n1. `Create 3 tasks: \\\"Setup project\\\", \\\"Write code\\\", \\\"Write tests\\\". Make them depend on each other in order.`\\n2. `List all tasks and show the dependency graph`\\n3. `Complete task 1 and then list tasks to see task 2 unblocked`\\n4. `Create a task board for refactoring: parse -> transform -> emit -> test, where transform and emit can run in parallel after parse`\\n\"\n },\n {\n \"version\": \"s08\",\n \"locale\": \"en\",\n \"title\": \"s08: Background Tasks\",\n \"content\": \"# s08: Background Tasks\\n\\n`s01 > s02 > s03 > s04 > s05 > s06 | s07 > [ s08 ] s09 > s10 > s11 > s12`\\n\\n> *\\\"Run slow operations in the background; the agent keeps thinking\\\"* -- daemon threads run commands, inject notifications on completion.\\n\\n## Problem\\n\\nSome commands take minutes: `npm install`, `pytest`, `docker build`. With a blocking loop, the model sits idle waiting. If the user asks \\\"install dependencies and while that runs, create the config file,\\\" the agent does them sequentially, not in parallel.\\n\\n## Solution\\n\\n```\\nMain thread Background thread\\n+-----------------+ +-----------------+\\n| agent loop | | subprocess runs |\\n| ... | | ... |\\n| [LLM call] <---+------- | enqueue(result) |\\n| ^drain queue | +-----------------+\\n+-----------------+\\n\\nTimeline:\\nAgent --[spawn A]--[spawn B]--[other work]----\\n | |\\n v v\\n [A runs] [B runs] (parallel)\\n | |\\n +-- results injected before next LLM call --+\\n```\\n\\n## How It Works\\n\\n1. BackgroundManager tracks tasks with a thread-safe notification queue.\\n\\n```python\\nclass BackgroundManager:\\n def __init__(self):\\n self.tasks = {}\\n self._notification_queue = []\\n self._lock = threading.Lock()\\n```\\n\\n2. `run()` starts a daemon thread and returns immediately.\\n\\n```python\\ndef run(self, command: str) -> str:\\n task_id = str(uuid.uuid4())[:8]\\n self.tasks[task_id] = {\\\"status\\\": \\\"running\\\", \\\"command\\\": command}\\n thread = threading.Thread(\\n target=self._execute, args=(task_id, command), daemon=True)\\n thread.start()\\n return f\\\"Background task {task_id} started\\\"\\n```\\n\\n3. When the subprocess finishes, its result goes into the notification queue.\\n\\n```python\\ndef _execute(self, task_id, command):\\n try:\\n r = subprocess.run(command, shell=True, cwd=WORKDIR,\\n capture_output=True, text=True, timeout=300)\\n output = (r.stdout + r.stderr).strip()[:50000]\\n except subprocess.TimeoutExpired:\\n output = \\\"Error: Timeout (300s)\\\"\\n with self._lock:\\n self._notification_queue.append({\\n \\\"task_id\\\": task_id, \\\"result\\\": output[:500]})\\n```\\n\\n4. The agent loop drains notifications before each LLM call.\\n\\n```python\\ndef agent_loop(messages: list):\\n while True:\\n notifs = BG.drain_notifications()\\n if notifs:\\n notif_text = \\\"\\\\n\\\".join(\\n f\\\"[bg:{n['task_id']}] {n['result']}\\\" for n in notifs)\\n messages.append({\\\"role\\\": \\\"user\\\",\\n \\\"content\\\": f\\\"\\\\n{notif_text}\\\\n\\\"\\n f\\\" \\\"})\\n messages.append({\\\"role\\\": \\\"assistant\\\",\\n \\\"content\\\": \\\"Noted background results.\\\"})\\n response = client.messages.create(...)\\n```\\n\\nThe loop stays single-threaded. Only subprocess I/O is parallelized.\\n\\n## What Changed From s07\\n\\n| Component | Before (s07) | After (s08) |\\n|----------------|------------------|----------------------------|\\n| Tools | 8 | 6 (base + background_run + check)|\\n| Execution | Blocking only | Blocking + background threads|\\n| Notification | None | Queue drained per loop |\\n| Concurrency | None | Daemon threads |\\n\\n## Try It\\n\\n```sh\\ncd learn-claude-code\\npython agents/s08_background_tasks.py\\n```\\n\\n1. `Run \\\"sleep 5 && echo done\\\" in the background, then create a file while it runs`\\n2. `Start 3 background tasks: \\\"sleep 2\\\", \\\"sleep 4\\\", \\\"sleep 6\\\". Check their status.`\\n3. `Run pytest in the background and keep working on other things`\\n\"\n },\n {\n \"version\": \"s09\",\n \"locale\": \"en\",\n \"title\": \"s09: Agent Teams\",\n \"content\": \"# s09: Agent Teams\\n\\n`s01 > s02 > s03 > s04 > s05 > s06 | s07 > s08 > [ s09 ] s10 > s11 > s12`\\n\\n> *\\\"When the task is too big for one, delegate to teammates\\\"* -- persistent teammates + async mailboxes.\\n\\n## Problem\\n\\nSubagents (s04) are disposable: spawn, work, return summary, die. No identity, no memory between invocations. Background tasks (s08) run shell commands but can't make LLM-guided decisions.\\n\\nReal teamwork needs: (1) persistent agents that outlive a single prompt, (2) identity and lifecycle management, (3) a communication channel between agents.\\n\\n## Solution\\n\\n```\\nTeammate lifecycle:\\n spawn -> WORKING -> IDLE -> WORKING -> ... -> SHUTDOWN\\n\\nCommunication:\\n .team/\\n config.json <- team roster + statuses\\n inbox/\\n alice.jsonl <- append-only, drain-on-read\\n bob.jsonl\\n lead.jsonl\\n\\n +--------+ send(\\\"alice\\\",\\\"bob\\\",\\\"...\\\") +--------+\\n | alice | -----------------------------> | bob |\\n | loop | bob.jsonl << {json_line} | loop |\\n +--------+ +--------+\\n ^ |\\n | BUS.read_inbox(\\\"alice\\\") |\\n +---- alice.jsonl -> read + drain ---------+\\n```\\n\\n## How It Works\\n\\n1. TeammateManager maintains config.json with the team roster.\\n\\n```python\\nclass TeammateManager:\\n def __init__(self, team_dir: Path):\\n self.dir = team_dir\\n self.dir.mkdir(exist_ok=True)\\n self.config_path = self.dir / \\\"config.json\\\"\\n self.config = self._load_config()\\n self.threads = {}\\n```\\n\\n2. `spawn()` creates a teammate and starts its agent loop in a thread.\\n\\n```python\\ndef spawn(self, name: str, role: str, prompt: str) -> str:\\n member = {\\\"name\\\": name, \\\"role\\\": role, \\\"status\\\": \\\"working\\\"}\\n self.config[\\\"members\\\"].append(member)\\n self._save_config()\\n thread = threading.Thread(\\n target=self._teammate_loop,\\n args=(name, role, prompt), daemon=True)\\n thread.start()\\n return f\\\"Spawned teammate '{name}' (role: {role})\\\"\\n```\\n\\n3. MessageBus: append-only JSONL inboxes. `send()` appends a JSON line; `read_inbox()` reads all and drains.\\n\\n```python\\nclass MessageBus:\\n def send(self, sender, to, content, msg_type=\\\"message\\\", extra=None):\\n msg = {\\\"type\\\": msg_type, \\\"from\\\": sender,\\n \\\"content\\\": content, \\\"timestamp\\\": time.time()}\\n if extra:\\n msg.update(extra)\\n with open(self.dir / f\\\"{to}.jsonl\\\", \\\"a\\\") as f:\\n f.write(json.dumps(msg) + \\\"\\\\n\\\")\\n\\n def read_inbox(self, name):\\n path = self.dir / f\\\"{name}.jsonl\\\"\\n if not path.exists(): return \\\"[]\\\"\\n msgs = [json.loads(l) for l in path.read_text().strip().splitlines() if l]\\n path.write_text(\\\"\\\") # drain\\n return json.dumps(msgs, indent=2)\\n```\\n\\n4. Each teammate checks its inbox before every LLM call, injecting received messages into context.\\n\\n```python\\ndef _teammate_loop(self, name, role, prompt):\\n messages = [{\\\"role\\\": \\\"user\\\", \\\"content\\\": prompt}]\\n for _ in range(50):\\n inbox = BUS.read_inbox(name)\\n if inbox != \\\"[]\\\":\\n messages.append({\\\"role\\\": \\\"user\\\",\\n \\\"content\\\": f\\\"{inbox} \\\"})\\n messages.append({\\\"role\\\": \\\"assistant\\\",\\n \\\"content\\\": \\\"Noted inbox messages.\\\"})\\n response = client.messages.create(...)\\n if response.stop_reason != \\\"tool_use\\\":\\n break\\n # execute tools, append results...\\n self._find_member(name)[\\\"status\\\"] = \\\"idle\\\"\\n```\\n\\n## What Changed From s08\\n\\n| Component | Before (s08) | After (s09) |\\n|----------------|------------------|----------------------------|\\n| Tools | 6 | 9 (+spawn/send/read_inbox) |\\n| Agents | Single | Lead + N teammates |\\n| Persistence | None | config.json + JSONL inboxes|\\n| Threads | Background cmds | Full agent loops per thread|\\n| Lifecycle | Fire-and-forget | idle -> working -> idle |\\n| Communication | None | message + broadcast |\\n\\n## Try It\\n\\n```sh\\ncd learn-claude-code\\npython agents/s09_agent_teams.py\\n```\\n\\n1. `Spawn alice (coder) and bob (tester). Have alice send bob a message.`\\n2. `Broadcast \\\"status update: phase 1 complete\\\" to all teammates`\\n3. `Check the lead inbox for any messages`\\n4. Type `/team` to see the team roster with statuses\\n5. Type `/inbox` to manually check the lead's inbox\\n\"\n },\n {\n \"version\": \"s10\",\n \"locale\": \"en\",\n \"title\": \"s10: Team Protocols\",\n \"content\": \"# s10: Team Protocols\\n\\n`s01 > s02 > s03 > s04 > s05 > s06 | s07 > s08 > s09 > [ s10 ] s11 > s12`\\n\\n> *\\\"Teammates need shared communication rules\\\"* -- one request-response pattern drives all negotiation.\\n\\n## Problem\\n\\nIn s09, teammates work and communicate but lack structured coordination:\\n\\n**Shutdown**: Killing a thread leaves files half-written and config.json stale. You need a handshake: the lead requests, the teammate approves (finish and exit) or rejects (keep working).\\n\\n**Plan approval**: When the lead says \\\"refactor the auth module,\\\" the teammate starts immediately. For high-risk changes, the lead should review the plan first.\\n\\nBoth share the same structure: one side sends a request with a unique ID, the other responds referencing that ID.\\n\\n## Solution\\n\\n```\\nShutdown Protocol Plan Approval Protocol\\n================== ======================\\n\\nLead Teammate Teammate Lead\\n | | | |\\n |--shutdown_req-->| |--plan_req------>|\\n | {req_id:\\\"abc\\\"} | | {req_id:\\\"xyz\\\"} |\\n | | | |\\n |<--shutdown_resp-| |<--plan_resp-----|\\n | {req_id:\\\"abc\\\", | | {req_id:\\\"xyz\\\", |\\n | approve:true} | | approve:true} |\\n\\nShared FSM:\\n [pending] --approve--> [approved]\\n [pending] --reject---> [rejected]\\n\\nTrackers:\\n shutdown_requests = {req_id: {target, status}}\\n plan_requests = {req_id: {from, plan, status}}\\n```\\n\\n## How It Works\\n\\n1. The lead initiates shutdown by generating a request_id and sending through the inbox.\\n\\n```python\\nshutdown_requests = {}\\n\\ndef handle_shutdown_request(teammate: str) -> str:\\n req_id = str(uuid.uuid4())[:8]\\n shutdown_requests[req_id] = {\\\"target\\\": teammate, \\\"status\\\": \\\"pending\\\"}\\n BUS.send(\\\"lead\\\", teammate, \\\"Please shut down gracefully.\\\",\\n \\\"shutdown_request\\\", {\\\"request_id\\\": req_id})\\n return f\\\"Shutdown request {req_id} sent (status: pending)\\\"\\n```\\n\\n2. The teammate receives the request and responds with approve/reject.\\n\\n```python\\nif tool_name == \\\"shutdown_response\\\":\\n req_id = args[\\\"request_id\\\"]\\n approve = args[\\\"approve\\\"]\\n shutdown_requests[req_id][\\\"status\\\"] = \\\"approved\\\" if approve else \\\"rejected\\\"\\n BUS.send(sender, \\\"lead\\\", args.get(\\\"reason\\\", \\\"\\\"),\\n \\\"shutdown_response\\\",\\n {\\\"request_id\\\": req_id, \\\"approve\\\": approve})\\n```\\n\\n3. Plan approval follows the identical pattern. The teammate submits a plan (generating a request_id), the lead reviews (referencing the same request_id).\\n\\n```python\\nplan_requests = {}\\n\\ndef handle_plan_review(request_id, approve, feedback=\\\"\\\"):\\n req = plan_requests[request_id]\\n req[\\\"status\\\"] = \\\"approved\\\" if approve else \\\"rejected\\\"\\n BUS.send(\\\"lead\\\", req[\\\"from\\\"], feedback,\\n \\\"plan_approval_response\\\",\\n {\\\"request_id\\\": request_id, \\\"approve\\\": approve})\\n```\\n\\nOne FSM, two applications. The same `pending -> approved | rejected` state machine handles any request-response protocol.\\n\\n## What Changed From s09\\n\\n| Component | Before (s09) | After (s10) |\\n|----------------|------------------|------------------------------|\\n| Tools | 9 | 12 (+shutdown_req/resp +plan)|\\n| Shutdown | Natural exit only| Request-response handshake |\\n| Plan gating | None | Submit/review with approval |\\n| Correlation | None | request_id per request |\\n| FSM | None | pending -> approved/rejected |\\n\\n## Try It\\n\\n```sh\\ncd learn-claude-code\\npython agents/s10_team_protocols.py\\n```\\n\\n1. `Spawn alice as a coder. Then request her shutdown.`\\n2. `List teammates to see alice's status after shutdown approval`\\n3. `Spawn bob with a risky refactoring task. Review and reject his plan.`\\n4. `Spawn charlie, have him submit a plan, then approve it.`\\n5. Type `/team` to monitor statuses\\n\"\n },\n {\n \"version\": \"s11\",\n \"locale\": \"en\",\n \"title\": \"s11: Autonomous Agents\",\n \"content\": \"# s11: Autonomous Agents\\n\\n`s01 > s02 > s03 > s04 > s05 > s06 | s07 > s08 > s09 > s10 > [ s11 ] s12`\\n\\n> *\\\"Teammates scan the board and claim tasks themselves\\\"* -- no need for the lead to assign each one.\\n\\n## Problem\\n\\nIn s09-s10, teammates only work when explicitly told to. The lead must spawn each one with a specific prompt. 10 unclaimed tasks on the board? The lead assigns each one manually. Doesn't scale.\\n\\nTrue autonomy: teammates scan the task board themselves, claim unclaimed tasks, work on them, then look for more.\\n\\nOne subtlety: after context compression (s06), the agent might forget who it is. Identity re-injection fixes this.\\n\\n## Solution\\n\\n```\\nTeammate lifecycle with idle cycle:\\n\\n+-------+\\n| spawn |\\n+---+---+\\n |\\n v\\n+-------+ tool_use +-------+\\n| WORK | <------------- | LLM |\\n+---+---+ +-------+\\n |\\n | stop_reason != tool_use (or idle tool called)\\n v\\n+--------+\\n| IDLE | poll every 5s for up to 60s\\n+---+----+\\n |\\n +---> check inbox --> message? ----------> WORK\\n |\\n +---> scan .tasks/ --> unclaimed? -------> claim -> WORK\\n |\\n +---> 60s timeout ----------------------> SHUTDOWN\\n\\nIdentity re-injection after compression:\\n if len(messages) <= 3:\\n messages.insert(0, identity_block)\\n```\\n\\n## How It Works\\n\\n1. The teammate loop has two phases: WORK and IDLE. When the LLM stops calling tools (or calls `idle`), the teammate enters IDLE.\\n\\n```python\\ndef _loop(self, name, role, prompt):\\n while True:\\n # -- WORK PHASE --\\n messages = [{\\\"role\\\": \\\"user\\\", \\\"content\\\": prompt}]\\n for _ in range(50):\\n response = client.messages.create(...)\\n if response.stop_reason != \\\"tool_use\\\":\\n break\\n # execute tools...\\n if idle_requested:\\n break\\n\\n # -- IDLE PHASE --\\n self._set_status(name, \\\"idle\\\")\\n resume = self._idle_poll(name, messages)\\n if not resume:\\n self._set_status(name, \\\"shutdown\\\")\\n return\\n self._set_status(name, \\\"working\\\")\\n```\\n\\n2. The idle phase polls inbox and task board in a loop.\\n\\n```python\\ndef _idle_poll(self, name, messages):\\n for _ in range(IDLE_TIMEOUT // POLL_INTERVAL): # 60s / 5s = 12\\n time.sleep(POLL_INTERVAL)\\n inbox = BUS.read_inbox(name)\\n if inbox:\\n messages.append({\\\"role\\\": \\\"user\\\",\\n \\\"content\\\": f\\\"{inbox} \\\"})\\n return True\\n unclaimed = scan_unclaimed_tasks()\\n if unclaimed:\\n claim_task(unclaimed[0][\\\"id\\\"], name)\\n messages.append({\\\"role\\\": \\\"user\\\",\\n \\\"content\\\": f\\\"Task #{unclaimed[0]['id']}: \\\"\\n f\\\"{unclaimed[0]['subject']} \\\"})\\n return True\\n return False # timeout -> shutdown\\n```\\n\\n3. Task board scanning: find pending, unowned, unblocked tasks.\\n\\n```python\\ndef scan_unclaimed_tasks() -> list:\\n unclaimed = []\\n for f in sorted(TASKS_DIR.glob(\\\"task_*.json\\\")):\\n task = json.loads(f.read_text())\\n if (task.get(\\\"status\\\") == \\\"pending\\\"\\n and not task.get(\\\"owner\\\")\\n and not task.get(\\\"blockedBy\\\")):\\n unclaimed.append(task)\\n return unclaimed\\n```\\n\\n4. Identity re-injection: when context is too short (compression happened), insert an identity block.\\n\\n```python\\nif len(messages) <= 3:\\n messages.insert(0, {\\\"role\\\": \\\"user\\\",\\n \\\"content\\\": f\\\"You are '{name}', role: {role}, \\\"\\n f\\\"team: {team_name}. Continue your work. \\\"})\\n messages.insert(1, {\\\"role\\\": \\\"assistant\\\",\\n \\\"content\\\": f\\\"I am {name}. Continuing.\\\"})\\n```\\n\\n## What Changed From s10\\n\\n| Component | Before (s10) | After (s11) |\\n|----------------|------------------|----------------------------|\\n| Tools | 12 | 14 (+idle, +claim_task) |\\n| Autonomy | Lead-directed | Self-organizing |\\n| Idle phase | None | Poll inbox + task board |\\n| Task claiming | Manual only | Auto-claim unclaimed tasks |\\n| Identity | System prompt | + re-injection after compress|\\n| Timeout | None | 60s idle -> auto shutdown |\\n\\n## Try It\\n\\n```sh\\ncd learn-claude-code\\npython agents/s11_autonomous_agents.py\\n```\\n\\n1. `Create 3 tasks on the board, then spawn alice and bob. Watch them auto-claim.`\\n2. `Spawn a coder teammate and let it find work from the task board itself`\\n3. `Create tasks with dependencies. Watch teammates respect the blocked order.`\\n4. Type `/tasks` to see the task board with owners\\n5. Type `/team` to monitor who is working vs idle\\n\"\n },\n {\n \"version\": \"s12\",\n \"locale\": \"en\",\n \"title\": \"s12: Worktree + Task Isolation\",\n \"content\": \"# s12: Worktree + Task Isolation\\n\\n`s01 > s02 > s03 > s04 > s05 > s06 | s07 > s08 > s09 > s10 > s11 > [ s12 ]`\\n\\n> *\\\"Each works in its own directory, no interference\\\"* -- tasks manage goals, worktrees manage directories, bound by ID.\\n\\n## Problem\\n\\nBy s11, agents can claim and complete tasks autonomously. But every task runs in one shared directory. Two agents refactoring different modules at the same time will collide: agent A edits `config.py`, agent B edits `config.py`, unstaged changes mix, and neither can roll back cleanly.\\n\\nThe task board tracks *what to do* but has no opinion about *where to do it*. The fix: give each task its own git worktree directory. Tasks manage goals, worktrees manage execution context. Bind them by task ID.\\n\\n## Solution\\n\\n```\\nControl plane (.tasks/) Execution plane (.worktrees/)\\n+------------------+ +------------------------+\\n| task_1.json | | auth-refactor/ |\\n| status: in_progress <------> branch: wt/auth-refactor\\n| worktree: \\\"auth-refactor\\\" | task_id: 1 |\\n+------------------+ +------------------------+\\n| task_2.json | | ui-login/ |\\n| status: pending <------> branch: wt/ui-login\\n| worktree: \\\"ui-login\\\" | task_id: 2 |\\n+------------------+ +------------------------+\\n |\\n index.json (worktree registry)\\n events.jsonl (lifecycle log)\\n\\nState machines:\\n Task: pending -> in_progress -> completed\\n Worktree: absent -> active -> removed | kept\\n```\\n\\n## How It Works\\n\\n1. **Create a task.** Persist the goal first.\\n\\n```python\\nTASKS.create(\\\"Implement auth refactor\\\")\\n# -> .tasks/task_1.json status=pending worktree=\\\"\\\"\\n```\\n\\n2. **Create a worktree and bind to the task.** Passing `task_id` auto-advances the task to `in_progress`.\\n\\n```python\\nWORKTREES.create(\\\"auth-refactor\\\", task_id=1)\\n# -> git worktree add -b wt/auth-refactor .worktrees/auth-refactor HEAD\\n# -> index.json gets new entry, task_1.json gets worktree=\\\"auth-refactor\\\"\\n```\\n\\nThe binding writes state to both sides:\\n\\n```python\\ndef bind_worktree(self, task_id, worktree):\\n task = self._load(task_id)\\n task[\\\"worktree\\\"] = worktree\\n if task[\\\"status\\\"] == \\\"pending\\\":\\n task[\\\"status\\\"] = \\\"in_progress\\\"\\n self._save(task)\\n```\\n\\n3. **Run commands in the worktree.** `cwd` points to the isolated directory.\\n\\n```python\\nsubprocess.run(command, shell=True, cwd=worktree_path,\\n capture_output=True, text=True, timeout=300)\\n```\\n\\n4. **Close out.** Two choices:\\n - `worktree_keep(name)` -- preserve the directory for later.\\n - `worktree_remove(name, complete_task=True)` -- remove directory, complete the bound task, emit event. One call handles teardown + completion.\\n\\n```python\\ndef remove(self, name, force=False, complete_task=False):\\n self._run_git([\\\"worktree\\\", \\\"remove\\\", wt[\\\"path\\\"]])\\n if complete_task and wt.get(\\\"task_id\\\") is not None:\\n self.tasks.update(wt[\\\"task_id\\\"], status=\\\"completed\\\")\\n self.tasks.unbind_worktree(wt[\\\"task_id\\\"])\\n self.events.emit(\\\"task.completed\\\", ...)\\n```\\n\\n5. **Event stream.** Every lifecycle step emits to `.worktrees/events.jsonl`:\\n\\n```json\\n{\\n \\\"event\\\": \\\"worktree.remove.after\\\",\\n \\\"task\\\": {\\\"id\\\": 1, \\\"status\\\": \\\"completed\\\"},\\n \\\"worktree\\\": {\\\"name\\\": \\\"auth-refactor\\\", \\\"status\\\": \\\"removed\\\"},\\n \\\"ts\\\": 1730000000\\n}\\n```\\n\\nEvents emitted: `worktree.create.before/after/failed`, `worktree.remove.before/after/failed`, `worktree.keep`, `task.completed`.\\n\\nAfter a crash, state reconstructs from `.tasks/` + `.worktrees/index.json` on disk. Conversation memory is volatile; file state is durable.\\n\\n## What Changed From s11\\n\\n| Component | Before (s11) | After (s12) |\\n|--------------------|----------------------------|----------------------------------------------|\\n| Coordination | Task board (owner/status) | Task board + explicit worktree binding |\\n| Execution scope | Shared directory | Task-scoped isolated directory |\\n| Recoverability | Task status only | Task status + worktree index |\\n| Teardown | Task completion | Task completion + explicit keep/remove |\\n| Lifecycle visibility | Implicit in logs | Explicit events in `.worktrees/events.jsonl` |\\n\\n## Try It\\n\\n```sh\\ncd learn-claude-code\\npython agents/s12_worktree_task_isolation.py\\n```\\n\\n1. `Create tasks for backend auth and frontend login page, then list tasks.`\\n2. `Create worktree \\\"auth-refactor\\\" for task 1, then bind task 2 to a new worktree \\\"ui-login\\\".`\\n3. `Run \\\"git status --short\\\" in worktree \\\"auth-refactor\\\".`\\n4. `Keep worktree \\\"ui-login\\\", then list worktrees and inspect events.`\\n5. `Remove worktree \\\"auth-refactor\\\" with complete_task=true, then list tasks/worktrees/events.`\\n\"\n },\n {\n \"version\": \"s01\",\n \"locale\": \"zh\",\n \"title\": \"s01: The Agent Loop (智能体循环)\",\n \"content\": \"# s01: The Agent Loop (智能体循环)\\n\\n`[ s01 ] s02 > s03 > s04 > s05 > s06 | s07 > s08 > s09 > s10 > s11 > s12`\\n\\n> *\\\"One loop & Bash is all you need\\\"* -- 一个工具 + 一个循环 = 一个智能体。\\n\\n## 问题\\n\\n语言模型能推理代码, 但碰不到真实世界 -- 不能读文件、跑测试、看报错。没有循环, 每次工具调用你都得手动把结果粘回去。你自己就是那个循环。\\n\\n## 解决方案\\n\\n```\\n+--------+ +-------+ +---------+\\n| User | ---> | LLM | ---> | Tool |\\n| prompt | | | | execute |\\n+--------+ +---+---+ +----+----+\\n ^ |\\n | tool_result |\\n +----------------+\\n (loop until stop_reason != \\\"tool_use\\\")\\n```\\n\\n一个退出条件控制整个流程。循环持续运行, 直到模型不再调用工具。\\n\\n## 工作原理\\n\\n1. 用户 prompt 作为第一条消息。\\n\\n```python\\nmessages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": query})\\n```\\n\\n2. 将消息和工具定义一起发给 LLM。\\n\\n```python\\nresponse = client.messages.create(\\n model=MODEL, system=SYSTEM, messages=messages,\\n tools=TOOLS, max_tokens=8000,\\n)\\n```\\n\\n3. 追加助手响应。检查 `stop_reason` -- 如果模型没有调用工具, 结束。\\n\\n```python\\nmessages.append({\\\"role\\\": \\\"assistant\\\", \\\"content\\\": response.content})\\nif response.stop_reason != \\\"tool_use\\\":\\n return\\n```\\n\\n4. 执行每个工具调用, 收集结果, 作为 user 消息追加。回到第 2 步。\\n\\n```python\\nresults = []\\nfor block in response.content:\\n if block.type == \\\"tool_use\\\":\\n output = run_bash(block.input[\\\"command\\\"])\\n results.append({\\n \\\"type\\\": \\\"tool_result\\\",\\n \\\"tool_use_id\\\": block.id,\\n \\\"content\\\": output,\\n })\\nmessages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": results})\\n```\\n\\n组装为一个完整函数:\\n\\n```python\\ndef agent_loop(query):\\n messages = [{\\\"role\\\": \\\"user\\\", \\\"content\\\": query}]\\n while True:\\n response = client.messages.create(\\n model=MODEL, system=SYSTEM, messages=messages,\\n tools=TOOLS, max_tokens=8000,\\n )\\n messages.append({\\\"role\\\": \\\"assistant\\\", \\\"content\\\": response.content})\\n\\n if response.stop_reason != \\\"tool_use\\\":\\n return\\n\\n results = []\\n for block in response.content:\\n if block.type == \\\"tool_use\\\":\\n output = run_bash(block.input[\\\"command\\\"])\\n results.append({\\n \\\"type\\\": \\\"tool_result\\\",\\n \\\"tool_use_id\\\": block.id,\\n \\\"content\\\": output,\\n })\\n messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": results})\\n```\\n\\n不到 30 行, 这就是整个智能体。后面 11 个章节都在这个循环上叠加机制 -- 循环本身始终不变。\\n\\n## 变更内容\\n\\n| 组件 | 之前 | 之后 |\\n|---------------|------------|--------------------------------|\\n| Agent loop | (无) | `while True` + stop_reason |\\n| Tools | (无) | `bash` (单一工具) |\\n| Messages | (无) | 累积式消息列表 |\\n| Control flow | (无) | `stop_reason != \\\"tool_use\\\"` |\\n\\n## 试一试\\n\\n```sh\\ncd learn-claude-code\\npython agents/s01_agent_loop.py\\n```\\n\\n试试这些 prompt (英文 prompt 对 LLM 效果更好, 也可以用中文):\\n\\n1. `Create a file called hello.py that prints \\\"Hello, World!\\\"`\\n2. `List all Python files in this directory`\\n3. `What is the current git branch?`\\n4. `Create a directory called test_output and write 3 files in it`\\n\"\n },\n {\n \"version\": \"s02\",\n \"locale\": \"zh\",\n \"title\": \"s02: Tool Use (工具使用)\",\n \"content\": \"# s02: Tool Use (工具使用)\\n\\n`s01 > [ s02 ] s03 > s04 > s05 > s06 | s07 > s08 > s09 > s10 > s11 > s12`\\n\\n> *\\\"加一个工具, 只加一个 handler\\\"* -- 循环不用动, 新工具注册进 dispatch map 就行。\\n\\n## 问题\\n\\n只有 `bash` 时, 所有操作都走 shell。`cat` 截断不可预测, `sed` 遇到特殊字符就崩, 每次 bash 调用都是不受约束的安全面。专用工具 (`read_file`, `write_file`) 可以在工具层面做路径沙箱。\\n\\n关键洞察: 加工具不需要改循环。\\n\\n## 解决方案\\n\\n```\\n+--------+ +-------+ +------------------+\\n| User | ---> | LLM | ---> | Tool Dispatch |\\n| prompt | | | | { |\\n+--------+ +---+---+ | bash: run_bash |\\n ^ | read: run_read |\\n | | write: run_wr |\\n +-----------+ edit: run_edit |\\n tool_result | } |\\n +------------------+\\n\\nThe dispatch map is a dict: {tool_name: handler_function}.\\nOne lookup replaces any if/elif chain.\\n```\\n\\n## 工作原理\\n\\n1. 每个工具有一个处理函数。路径沙箱防止逃逸工作区。\\n\\n```python\\ndef safe_path(p: str) -> Path:\\n path = (WORKDIR / p).resolve()\\n if not path.is_relative_to(WORKDIR):\\n raise ValueError(f\\\"Path escapes workspace: {p}\\\")\\n return path\\n\\ndef run_read(path: str, limit: int = None) -> str:\\n text = safe_path(path).read_text()\\n lines = text.splitlines()\\n if limit and limit < len(lines):\\n lines = lines[:limit]\\n return \\\"\\\\n\\\".join(lines)[:50000]\\n```\\n\\n2. dispatch map 将工具名映射到处理函数。\\n\\n```python\\nTOOL_HANDLERS = {\\n \\\"bash\\\": lambda **kw: run_bash(kw[\\\"command\\\"]),\\n \\\"read_file\\\": lambda **kw: run_read(kw[\\\"path\\\"], kw.get(\\\"limit\\\")),\\n \\\"write_file\\\": lambda **kw: run_write(kw[\\\"path\\\"], kw[\\\"content\\\"]),\\n \\\"edit_file\\\": lambda **kw: run_edit(kw[\\\"path\\\"], kw[\\\"old_text\\\"],\\n kw[\\\"new_text\\\"]),\\n}\\n```\\n\\n3. 循环中按名称查找处理函数。循环体本身与 s01 完全一致。\\n\\n```python\\nfor block in response.content:\\n if block.type == \\\"tool_use\\\":\\n handler = TOOL_HANDLERS.get(block.name)\\n output = handler(**block.input) if handler \\\\\\n else f\\\"Unknown tool: {block.name}\\\"\\n results.append({\\n \\\"type\\\": \\\"tool_result\\\",\\n \\\"tool_use_id\\\": block.id,\\n \\\"content\\\": output,\\n })\\n```\\n\\n加工具 = 加 handler + 加 schema。循环永远不变。\\n\\n## 相对 s01 的变更\\n\\n| 组件 | 之前 (s01) | 之后 (s02) |\\n|----------------|--------------------|--------------------------------|\\n| Tools | 1 (仅 bash) | 4 (bash, read, write, edit) |\\n| Dispatch | 硬编码 bash 调用 | `TOOL_HANDLERS` 字典 |\\n| 路径安全 | 无 | `safe_path()` 沙箱 |\\n| Agent loop | 不变 | 不变 |\\n\\n## 试一试\\n\\n```sh\\ncd learn-claude-code\\npython agents/s02_tool_use.py\\n```\\n\\n试试这些 prompt (英文 prompt 对 LLM 效果更好, 也可以用中文):\\n\\n1. `Read the file requirements.txt`\\n2. `Create a file called greet.py with a greet(name) function`\\n3. `Edit greet.py to add a docstring to the function`\\n4. `Read greet.py to verify the edit worked`\\n\"\n },\n {\n \"version\": \"s03\",\n \"locale\": \"zh\",\n \"title\": \"s03: TodoWrite (待办写入)\",\n \"content\": \"# s03: TodoWrite (待办写入)\\n\\n`s01 > s02 > [ s03 ] s04 > s05 > s06 | s07 > s08 > s09 > s10 > s11 > s12`\\n\\n> *\\\"没有计划的 agent 走哪算哪\\\"* -- 先列步骤再动手, 完成率翻倍。\\n\\n## 问题\\n\\n多步任务中, 模型会丢失进度 -- 重复做过的事、跳步、跑偏。对话越长越严重: 工具结果不断填满上下文, 系统提示的影响力逐渐被稀释。一个 10 步重构可能做完 1-3 步就开始即兴发挥, 因为 4-10 步已经被挤出注意力了。\\n\\n## 解决方案\\n\\n```\\n+--------+ +-------+ +---------+\\n| User | ---> | LLM | ---> | Tools |\\n| prompt | | | | + todo |\\n+--------+ +---+---+ +----+----+\\n ^ |\\n | tool_result |\\n +----------------+\\n |\\n +-----------+-----------+\\n | TodoManager state |\\n | [ ] task A |\\n | [>] task B <- doing |\\n | [x] task C |\\n +-----------------------+\\n |\\n if rounds_since_todo >= 3:\\n inject into tool_result\\n```\\n\\n## 工作原理\\n\\n1. TodoManager 存储带状态的项目。同一时间只允许一个 `in_progress`。\\n\\n```python\\nclass TodoManager:\\n def update(self, items: list) -> str:\\n validated, in_progress_count = [], 0\\n for item in items:\\n status = item.get(\\\"status\\\", \\\"pending\\\")\\n if status == \\\"in_progress\\\":\\n in_progress_count += 1\\n validated.append({\\\"id\\\": item[\\\"id\\\"], \\\"text\\\": item[\\\"text\\\"],\\n \\\"status\\\": status})\\n if in_progress_count > 1:\\n raise ValueError(\\\"Only one task can be in_progress\\\")\\n self.items = validated\\n return self.render()\\n```\\n\\n2. `todo` 工具和其他工具一样加入 dispatch map。\\n\\n```python\\nTOOL_HANDLERS = {\\n # ...base tools...\\n \\\"todo\\\": lambda **kw: TODO.update(kw[\\\"items\\\"]),\\n}\\n```\\n\\n3. nag reminder: 模型连续 3 轮以上不调用 `todo` 时注入提醒。\\n\\n```python\\nif rounds_since_todo >= 3 and messages:\\n last = messages[-1]\\n if last[\\\"role\\\"] == \\\"user\\\" and isinstance(last.get(\\\"content\\\"), list):\\n last[\\\"content\\\"].insert(0, {\\n \\\"type\\\": \\\"text\\\",\\n \\\"text\\\": \\\"Update your todos. \\\",\\n })\\n```\\n\\n\\\"同时只能有一个 in_progress\\\" 强制顺序聚焦。nag reminder 制造问责压力 -- 你不更新计划, 系统就追着你问。\\n\\n## 相对 s02 的变更\\n\\n| 组件 | 之前 (s02) | 之后 (s03) |\\n|----------------|------------------|--------------------------------|\\n| Tools | 4 | 5 (+todo) |\\n| 规划 | 无 | 带状态的 TodoManager |\\n| Nag 注入 | 无 | 3 轮后注入 `` |\\n| Agent loop | 简单分发 | + rounds_since_todo 计数器 |\\n\\n## 试一试\\n\\n```sh\\ncd learn-claude-code\\npython agents/s03_todo_write.py\\n```\\n\\n试试这些 prompt (英文 prompt 对 LLM 效果更好, 也可以用中文):\\n\\n1. `Refactor the file hello.py: add type hints, docstrings, and a main guard`\\n2. `Create a Python package with __init__.py, utils.py, and tests/test_utils.py`\\n3. `Review all Python files and fix any style issues`\\n\"\n },\n {\n \"version\": \"s04\",\n \"locale\": \"zh\",\n \"title\": \"s04: Subagents (子智能体)\",\n \"content\": \"# s04: Subagents (子智能体)\\n\\n`s01 > s02 > s03 > [ s04 ] s05 > s06 | s07 > s08 > s09 > s10 > s11 > s12`\\n\\n> *\\\"大任务拆小, 每个小任务干净的上下文\\\"* -- 子智能体用独立 messages[], 不污染主对话。\\n\\n## 问题\\n\\n智能体工作越久, messages 数组越胖。每次读文件、跑命令的输出都永久留在上下文里。\\\"这个项目用什么测试框架?\\\" 可能要读 5 个文件, 但父智能体只需要一个词: \\\"pytest。\\\"\\n\\n## 解决方案\\n\\n```\\nParent agent Subagent\\n+------------------+ +------------------+\\n| messages=[...] | | messages=[] | <-- fresh\\n| | dispatch | |\\n| tool: task | ----------> | while tool_use: |\\n| prompt=\\\"...\\\" | | call tools |\\n| | summary | append results |\\n| result = \\\"...\\\" | <---------- | return last text |\\n+------------------+ +------------------+\\n\\nParent context stays clean. Subagent context is discarded.\\n```\\n\\n## 工作原理\\n\\n1. 父智能体有一个 `task` 工具。子智能体拥有除 `task` 外的所有基础工具 (禁止递归生成)。\\n\\n```python\\nPARENT_TOOLS = CHILD_TOOLS + [\\n {\\\"name\\\": \\\"task\\\",\\n \\\"description\\\": \\\"Spawn a subagent with fresh context.\\\",\\n \\\"input_schema\\\": {\\n \\\"type\\\": \\\"object\\\",\\n \\\"properties\\\": {\\\"prompt\\\": {\\\"type\\\": \\\"string\\\"}},\\n \\\"required\\\": [\\\"prompt\\\"],\\n }},\\n]\\n```\\n\\n2. 子智能体以 `messages=[]` 启动, 运行自己的循环。只有最终文本返回给父智能体。\\n\\n```python\\ndef run_subagent(prompt: str) -> str:\\n sub_messages = [{\\\"role\\\": \\\"user\\\", \\\"content\\\": prompt}]\\n for _ in range(30): # safety limit\\n response = client.messages.create(\\n model=MODEL, system=SUBAGENT_SYSTEM,\\n messages=sub_messages,\\n tools=CHILD_TOOLS, max_tokens=8000,\\n )\\n sub_messages.append({\\\"role\\\": \\\"assistant\\\",\\n \\\"content\\\": response.content})\\n if response.stop_reason != \\\"tool_use\\\":\\n break\\n results = []\\n for block in response.content:\\n if block.type == \\\"tool_use\\\":\\n handler = TOOL_HANDLERS.get(block.name)\\n output = handler(**block.input)\\n results.append({\\\"type\\\": \\\"tool_result\\\",\\n \\\"tool_use_id\\\": block.id,\\n \\\"content\\\": str(output)[:50000]})\\n sub_messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": results})\\n return \\\"\\\".join(\\n b.text for b in response.content if hasattr(b, \\\"text\\\")\\n ) or \\\"(no summary)\\\"\\n```\\n\\n子智能体可能跑了 30+ 次工具调用, 但整个消息历史直接丢弃。父智能体收到的只是一段摘要文本, 作为普通 `tool_result` 返回。\\n\\n## 相对 s03 的变更\\n\\n| 组件 | 之前 (s03) | 之后 (s04) |\\n|----------------|------------------|-------------------------------|\\n| Tools | 5 | 5 (基础) + task (仅父端) |\\n| 上下文 | 单一共享 | 父 + 子隔离 |\\n| Subagent | 无 | `run_subagent()` 函数 |\\n| 返回值 | 不适用 | 仅摘要文本 |\\n\\n## 试一试\\n\\n```sh\\ncd learn-claude-code\\npython agents/s04_subagent.py\\n```\\n\\n试试这些 prompt (英文 prompt 对 LLM 效果更好, 也可以用中文):\\n\\n1. `Use a subtask to find what testing framework this project uses`\\n2. `Delegate: read all .py files and summarize what each one does`\\n3. `Use a task to create a new module, then verify it from here`\\n\"\n },\n {\n \"version\": \"s05\",\n \"locale\": \"zh\",\n \"title\": \"s05: Skills (技能加载)\",\n \"content\": \"# s05: Skills (技能加载)\\n\\n`s01 > s02 > s03 > s04 > [ s05 ] s06 | s07 > s08 > s09 > s10 > s11 > s12`\\n\\n> *\\\"用到什么知识, 临时加载什么知识\\\"* -- 通过 tool_result 注入, 不塞 system prompt。\\n\\n## 问题\\n\\n你希望智能体遵循特定领域的工作流: git 约定、测试模式、代码审查清单。全塞进系统提示太浪费 -- 10 个技能, 每个 2000 token, 就是 20,000 token, 大部分跟当前任务毫无关系。\\n\\n## 解决方案\\n\\n```\\nSystem prompt (Layer 1 -- always present):\\n+--------------------------------------+\\n| You are a coding agent. |\\n| Skills available: |\\n| - git: Git workflow helpers | ~100 tokens/skill\\n| - test: Testing best practices |\\n+--------------------------------------+\\n\\nWhen model calls load_skill(\\\"git\\\"):\\n+--------------------------------------+\\n| tool_result (Layer 2 -- on demand): |\\n| |\\n| Full git workflow instructions... | ~2000 tokens\\n| Step 1: ... |\\n| |\\n+--------------------------------------+\\n```\\n\\n第一层: 系统提示中放技能名称 (低成本)。第二层: tool_result 中按需放完整内容。\\n\\n## 工作原理\\n\\n1. 每个技能是一个目录, 包含 `SKILL.md` 文件和 YAML frontmatter。\\n\\n```\\nskills/\\n pdf/\\n SKILL.md # ---\\\\n name: pdf\\\\n description: Process PDF files\\\\n ---\\\\n ...\\n code-review/\\n SKILL.md # ---\\\\n name: code-review\\\\n description: Review code\\\\n ---\\\\n ...\\n```\\n\\n2. SkillLoader 递归扫描 `SKILL.md` 文件, 用目录名作为技能标识。\\n\\n```python\\nclass SkillLoader:\\n def __init__(self, skills_dir: Path):\\n self.skills = {}\\n for f in sorted(skills_dir.rglob(\\\"SKILL.md\\\")):\\n text = f.read_text()\\n meta, body = self._parse_frontmatter(text)\\n name = meta.get(\\\"name\\\", f.parent.name)\\n self.skills[name] = {\\\"meta\\\": meta, \\\"body\\\": body}\\n\\n def get_descriptions(self) -> str:\\n lines = []\\n for name, skill in self.skills.items():\\n desc = skill[\\\"meta\\\"].get(\\\"description\\\", \\\"\\\")\\n lines.append(f\\\" - {name}: {desc}\\\")\\n return \\\"\\\\n\\\".join(lines)\\n\\n def get_content(self, name: str) -> str:\\n skill = self.skills.get(name)\\n if not skill:\\n return f\\\"Error: Unknown skill '{name}'.\\\"\\n return f\\\"\\\\n{skill['body']}\\\\n \\\"\\n```\\n\\n3. 第一层写入系统提示。第二层不过是 dispatch map 中的又一个工具。\\n\\n```python\\nSYSTEM = f\\\"\\\"\\\"You are a coding agent at {WORKDIR}.\\nSkills available:\\n{SKILL_LOADER.get_descriptions()}\\\"\\\"\\\"\\n\\nTOOL_HANDLERS = {\\n # ...base tools...\\n \\\"load_skill\\\": lambda **kw: SKILL_LOADER.get_content(kw[\\\"name\\\"]),\\n}\\n```\\n\\n模型知道有哪些技能 (便宜), 需要时再加载完整内容 (贵)。\\n\\n## 相对 s04 的变更\\n\\n| 组件 | 之前 (s04) | 之后 (s05) |\\n|----------------|------------------|--------------------------------|\\n| Tools | 5 (基础 + task) | 5 (基础 + load_skill) |\\n| 系统提示 | 静态字符串 | + 技能描述列表 |\\n| 知识库 | 无 | skills/\\\\*/SKILL.md 文件 |\\n| 注入方式 | 无 | 两层 (系统提示 + result) |\\n\\n## 试一试\\n\\n```sh\\ncd learn-claude-code\\npython agents/s05_skill_loading.py\\n```\\n\\n试试这些 prompt (英文 prompt 对 LLM 效果更好, 也可以用中文):\\n\\n1. `What skills are available?`\\n2. `Load the agent-builder skill and follow its instructions`\\n3. `I need to do a code review -- load the relevant skill first`\\n4. `Build an MCP server using the mcp-builder skill`\\n\"\n },\n {\n \"version\": \"s06\",\n \"locale\": \"zh\",\n \"title\": \"s06: Context Compact (上下文压缩)\",\n \"content\": \"# s06: Context Compact (上下文压缩)\\n\\n`s01 > s02 > s03 > s04 > s05 > [ s06 ] | s07 > s08 > s09 > s10 > s11 > s12`\\n\\n> *\\\"上下文总会满, 要有办法腾地方\\\"* -- 三层压缩策略, 换来无限会话。\\n\\n## 问题\\n\\n上下文窗口是有限的。读一个 1000 行的文件就吃掉 ~4000 token; 读 30 个文件、跑 20 条命令, 轻松突破 100k token。不压缩, 智能体根本没法在大项目里干活。\\n\\n## 解决方案\\n\\n三层压缩, 激进程度递增:\\n\\n```\\nEvery turn:\\n+------------------+\\n| Tool call result |\\n+------------------+\\n |\\n v\\n[Layer 1: micro_compact] (silent, every turn)\\n Replace tool_result > 3 turns old\\n with \\\"[Previous: used {tool_name}]\\\"\\n |\\n v\\n[Check: tokens > 50000?]\\n | |\\n no yes\\n | |\\n v v\\ncontinue [Layer 2: auto_compact]\\n Save transcript to .transcripts/\\n LLM summarizes conversation.\\n Replace all messages with [summary].\\n |\\n v\\n [Layer 3: compact tool]\\n Model calls compact explicitly.\\n Same summarization as auto_compact.\\n```\\n\\n## 工作原理\\n\\n1. **第一层 -- micro_compact**: 每次 LLM 调用前, 将旧的 tool result 替换为占位符。\\n\\n```python\\ndef micro_compact(messages: list) -> list:\\n tool_results = []\\n for i, msg in enumerate(messages):\\n if msg[\\\"role\\\"] == \\\"user\\\" and isinstance(msg.get(\\\"content\\\"), list):\\n for j, part in enumerate(msg[\\\"content\\\"]):\\n if isinstance(part, dict) and part.get(\\\"type\\\") == \\\"tool_result\\\":\\n tool_results.append((i, j, part))\\n if len(tool_results) <= KEEP_RECENT:\\n return messages\\n for _, _, part in tool_results[:-KEEP_RECENT]:\\n if len(part.get(\\\"content\\\", \\\"\\\")) > 100:\\n part[\\\"content\\\"] = f\\\"[Previous: used {tool_name}]\\\"\\n return messages\\n```\\n\\n2. **第二层 -- auto_compact**: token 超过阈值时, 保存完整对话到磁盘, 让 LLM 做摘要。\\n\\n```python\\ndef auto_compact(messages: list) -> list:\\n # Save transcript for recovery\\n transcript_path = TRANSCRIPT_DIR / f\\\"transcript_{int(time.time())}.jsonl\\\"\\n with open(transcript_path, \\\"w\\\") as f:\\n for msg in messages:\\n f.write(json.dumps(msg, default=str) + \\\"\\\\n\\\")\\n # LLM summarizes\\n response = client.messages.create(\\n model=MODEL,\\n messages=[{\\\"role\\\": \\\"user\\\", \\\"content\\\":\\n \\\"Summarize this conversation for continuity...\\\"\\n + json.dumps(messages, default=str)[:80000]}],\\n max_tokens=2000,\\n )\\n return [\\n {\\\"role\\\": \\\"user\\\", \\\"content\\\": f\\\"[Compressed]\\\\n\\\\n{response.content[0].text}\\\"},\\n {\\\"role\\\": \\\"assistant\\\", \\\"content\\\": \\\"Understood. Continuing.\\\"},\\n ]\\n```\\n\\n3. **第三层 -- manual compact**: `compact` 工具按需触发同样的摘要机制。\\n\\n4. 循环整合三层:\\n\\n```python\\ndef agent_loop(messages: list):\\n while True:\\n micro_compact(messages) # Layer 1\\n if estimate_tokens(messages) > THRESHOLD:\\n messages[:] = auto_compact(messages) # Layer 2\\n response = client.messages.create(...)\\n # ... tool execution ...\\n if manual_compact:\\n messages[:] = auto_compact(messages) # Layer 3\\n```\\n\\n完整历史通过 transcript 保存在磁盘上。信息没有真正丢失, 只是移出了活跃上下文。\\n\\n## 相对 s05 的变更\\n\\n| 组件 | 之前 (s05) | 之后 (s06) |\\n|----------------|------------------|--------------------------------|\\n| Tools | 5 | 5 (基础 + compact) |\\n| 上下文管理 | 无 | 三层压缩 |\\n| Micro-compact | 无 | 旧结果 -> 占位符 |\\n| Auto-compact | 无 | token 阈值触发 |\\n| Transcripts | 无 | 保存到 .transcripts/ |\\n\\n## 试一试\\n\\n```sh\\ncd learn-claude-code\\npython agents/s06_context_compact.py\\n```\\n\\n试试这些 prompt (英文 prompt 对 LLM 效果更好, 也可以用中文):\\n\\n1. `Read every Python file in the agents/ directory one by one` (观察 micro-compact 替换旧结果)\\n2. `Keep reading files until compression triggers automatically`\\n3. `Use the compact tool to manually compress the conversation`\\n\"\n },\n {\n \"version\": \"s07\",\n \"locale\": \"zh\",\n \"title\": \"s07: Task System (任务系统)\",\n \"content\": \"# s07: Task System (任务系统)\\n\\n`s01 > s02 > s03 > s04 > s05 > s06 | [ s07 ] s08 > s09 > s10 > s11 > s12`\\n\\n> *\\\"大目标要拆成小任务, 排好序, 记在磁盘上\\\"* -- 文件持久化的任务图, 为多 agent 协作打基础。\\n\\n## 问题\\n\\ns03 的 TodoManager 只是内存中的扁平清单: 没有顺序、没有依赖、状态只有做完没做完。真实目标是有结构的 -- 任务 B 依赖任务 A, 任务 C 和 D 可以并行, 任务 E 要等 C 和 D 都完成。\\n\\n没有显式的关系, 智能体分不清什么能做、什么被卡住、什么能同时跑。而且清单只活在内存里, 上下文压缩 (s06) 一跑就没了。\\n\\n## 解决方案\\n\\n把扁平清单升级为持久化到磁盘的**任务图**。每个任务是一个 JSON 文件, 有状态、前置依赖 (`blockedBy`) 和后置依赖 (`blocks`)。任务图随时回答三个问题:\\n\\n- **什么可以做?** -- 状态为 `pending` 且 `blockedBy` 为空的任务。\\n- **什么被卡住?** -- 等待前置任务完成的任务。\\n- **什么做完了?** -- 状态为 `completed` 的任务, 完成时自动解锁后续任务。\\n\\n```\\n.tasks/\\n task_1.json {\\\"id\\\":1, \\\"status\\\":\\\"completed\\\"}\\n task_2.json {\\\"id\\\":2, \\\"blockedBy\\\":[1], \\\"status\\\":\\\"pending\\\"}\\n task_3.json {\\\"id\\\":3, \\\"blockedBy\\\":[1], \\\"status\\\":\\\"pending\\\"}\\n task_4.json {\\\"id\\\":4, \\\"blockedBy\\\":[2,3], \\\"status\\\":\\\"pending\\\"}\\n\\n任务图 (DAG):\\n +----------+\\n +--> | task 2 | --+\\n | | pending | |\\n+----------+ +----------+ +--> +----------+\\n| task 1 | | task 4 |\\n| completed| --> +----------+ +--> | blocked |\\n+----------+ | task 3 | --+ +----------+\\n | pending |\\n +----------+\\n\\n顺序: task 1 必须先完成, 才能开始 2 和 3\\n并行: task 2 和 3 可以同时执行\\n依赖: task 4 要等 2 和 3 都完成\\n状态: pending -> in_progress -> completed\\n```\\n\\n这个任务图是 s07 之后所有机制的协调骨架: 后台执行 (s08)、多 agent 团队 (s09+)、worktree 隔离 (s12) 都读写这同一个结构。\\n\\n## 工作原理\\n\\n1. **TaskManager**: 每个任务一个 JSON 文件, CRUD + 依赖图。\\n\\n```python\\nclass TaskManager:\\n def __init__(self, tasks_dir: Path):\\n self.dir = tasks_dir\\n self.dir.mkdir(exist_ok=True)\\n self._next_id = self._max_id() + 1\\n\\n def create(self, subject, description=\\\"\\\"):\\n task = {\\\"id\\\": self._next_id, \\\"subject\\\": subject,\\n \\\"status\\\": \\\"pending\\\", \\\"blockedBy\\\": [],\\n \\\"blocks\\\": [], \\\"owner\\\": \\\"\\\"}\\n self._save(task)\\n self._next_id += 1\\n return json.dumps(task, indent=2)\\n```\\n\\n2. **依赖解除**: 完成任务时, 自动将其 ID 从其他任务的 `blockedBy` 中移除, 解锁后续任务。\\n\\n```python\\ndef _clear_dependency(self, completed_id):\\n for f in self.dir.glob(\\\"task_*.json\\\"):\\n task = json.loads(f.read_text())\\n if completed_id in task.get(\\\"blockedBy\\\", []):\\n task[\\\"blockedBy\\\"].remove(completed_id)\\n self._save(task)\\n```\\n\\n3. **状态变更 + 依赖关联**: `update` 处理状态转换和依赖边。\\n\\n```python\\ndef update(self, task_id, status=None,\\n add_blocked_by=None, add_blocks=None):\\n task = self._load(task_id)\\n if status:\\n task[\\\"status\\\"] = status\\n if status == \\\"completed\\\":\\n self._clear_dependency(task_id)\\n self._save(task)\\n```\\n\\n4. 四个任务工具加入 dispatch map。\\n\\n```python\\nTOOL_HANDLERS = {\\n # ...base tools...\\n \\\"task_create\\\": lambda **kw: TASKS.create(kw[\\\"subject\\\"]),\\n \\\"task_update\\\": lambda **kw: TASKS.update(kw[\\\"task_id\\\"], kw.get(\\\"status\\\")),\\n \\\"task_list\\\": lambda **kw: TASKS.list_all(),\\n \\\"task_get\\\": lambda **kw: TASKS.get(kw[\\\"task_id\\\"]),\\n}\\n```\\n\\n从 s07 起, 任务图是多步工作的默认选择。s03 的 Todo 仍可用于单次会话内的快速清单。\\n\\n## 相对 s06 的变更\\n\\n| 组件 | 之前 (s06) | 之后 (s07) |\\n|---|---|---|\\n| Tools | 5 | 8 (`task_create/update/list/get`) |\\n| 规划模型 | 扁平清单 (仅内存) | 带依赖关系的任务图 (磁盘) |\\n| 关系 | 无 | `blockedBy` + `blocks` 边 |\\n| 状态追踪 | 做完没做完 | `pending` -> `in_progress` -> `completed` |\\n| 持久化 | 压缩后丢失 | 压缩和重启后存活 |\\n\\n## 试一试\\n\\n```sh\\ncd learn-claude-code\\npython agents/s07_task_system.py\\n```\\n\\n试试这些 prompt (英文 prompt 对 LLM 效果更好, 也可以用中文):\\n\\n1. `Create 3 tasks: \\\"Setup project\\\", \\\"Write code\\\", \\\"Write tests\\\". Make them depend on each other in order.`\\n2. `List all tasks and show the dependency graph`\\n3. `Complete task 1 and then list tasks to see task 2 unblocked`\\n4. `Create a task board for refactoring: parse -> transform -> emit -> test, where transform and emit can run in parallel after parse`\\n\"\n },\n {\n \"version\": \"s08\",\n \"locale\": \"zh\",\n \"title\": \"s08: Background Tasks (后台任务)\",\n \"content\": \"# s08: Background Tasks (后台任务)\\n\\n`s01 > s02 > s03 > s04 > s05 > s06 | s07 > [ s08 ] s09 > s10 > s11 > s12`\\n\\n> *\\\"慢操作丢后台, agent 继续想下一步\\\"* -- 后台线程跑命令, 完成后注入通知。\\n\\n## 问题\\n\\n有些命令要跑好几分钟: `npm install`、`pytest`、`docker build`。阻塞式循环下模型只能干等。用户说 \\\"装依赖, 顺便建个配置文件\\\", 智能体却只能一个一个来。\\n\\n## 解决方案\\n\\n```\\nMain thread Background thread\\n+-----------------+ +-----------------+\\n| agent loop | | subprocess runs |\\n| ... | | ... |\\n| [LLM call] <---+------- | enqueue(result) |\\n| ^drain queue | +-----------------+\\n+-----------------+\\n\\nTimeline:\\nAgent --[spawn A]--[spawn B]--[other work]----\\n | |\\n v v\\n [A runs] [B runs] (parallel)\\n | |\\n +-- results injected before next LLM call --+\\n```\\n\\n## 工作原理\\n\\n1. BackgroundManager 用线程安全的通知队列追踪任务。\\n\\n```python\\nclass BackgroundManager:\\n def __init__(self):\\n self.tasks = {}\\n self._notification_queue = []\\n self._lock = threading.Lock()\\n```\\n\\n2. `run()` 启动守护线程, 立即返回。\\n\\n```python\\ndef run(self, command: str) -> str:\\n task_id = str(uuid.uuid4())[:8]\\n self.tasks[task_id] = {\\\"status\\\": \\\"running\\\", \\\"command\\\": command}\\n thread = threading.Thread(\\n target=self._execute, args=(task_id, command), daemon=True)\\n thread.start()\\n return f\\\"Background task {task_id} started\\\"\\n```\\n\\n3. 子进程完成后, 结果进入通知队列。\\n\\n```python\\ndef _execute(self, task_id, command):\\n try:\\n r = subprocess.run(command, shell=True, cwd=WORKDIR,\\n capture_output=True, text=True, timeout=300)\\n output = (r.stdout + r.stderr).strip()[:50000]\\n except subprocess.TimeoutExpired:\\n output = \\\"Error: Timeout (300s)\\\"\\n with self._lock:\\n self._notification_queue.append({\\n \\\"task_id\\\": task_id, \\\"result\\\": output[:500]})\\n```\\n\\n4. 每次 LLM 调用前排空通知队列。\\n\\n```python\\ndef agent_loop(messages: list):\\n while True:\\n notifs = BG.drain_notifications()\\n if notifs:\\n notif_text = \\\"\\\\n\\\".join(\\n f\\\"[bg:{n['task_id']}] {n['result']}\\\" for n in notifs)\\n messages.append({\\\"role\\\": \\\"user\\\",\\n \\\"content\\\": f\\\"\\\\n{notif_text}\\\\n\\\"\\n f\\\" \\\"})\\n messages.append({\\\"role\\\": \\\"assistant\\\",\\n \\\"content\\\": \\\"Noted background results.\\\"})\\n response = client.messages.create(...)\\n```\\n\\n循环保持单线程。只有子进程 I/O 被并行化。\\n\\n## 相对 s07 的变更\\n\\n| 组件 | 之前 (s07) | 之后 (s08) |\\n|----------------|------------------|------------------------------------|\\n| Tools | 8 | 6 (基础 + background_run + check) |\\n| 执行方式 | 仅阻塞 | 阻塞 + 后台线程 |\\n| 通知机制 | 无 | 每轮排空的队列 |\\n| 并发 | 无 | 守护线程 |\\n\\n## 试一试\\n\\n```sh\\ncd learn-claude-code\\npython agents/s08_background_tasks.py\\n```\\n\\n试试这些 prompt (英文 prompt 对 LLM 效果更好, 也可以用中文):\\n\\n1. `Run \\\"sleep 5 && echo done\\\" in the background, then create a file while it runs`\\n2. `Start 3 background tasks: \\\"sleep 2\\\", \\\"sleep 4\\\", \\\"sleep 6\\\". Check their status.`\\n3. `Run pytest in the background and keep working on other things`\\n\"\n },\n {\n \"version\": \"s09\",\n \"locale\": \"zh\",\n \"title\": \"s09: Agent Teams (智能体团队)\",\n \"content\": \"# s09: Agent Teams (智能体团队)\\n\\n`s01 > s02 > s03 > s04 > s05 > s06 | s07 > s08 > [ s09 ] s10 > s11 > s12`\\n\\n> *\\\"任务太大一个人干不完, 要能分给队友\\\"* -- 持久化队友 + JSONL 邮箱。\\n\\n## 问题\\n\\n子智能体 (s04) 是一次性的: 生成、干活、返回摘要、消亡。没有身份, 没有跨调用的记忆。后台任务 (s08) 能跑 shell 命令, 但做不了 LLM 引导的决策。\\n\\n真正的团队协作需要三样东西: (1) 能跨多轮对话存活的持久智能体, (2) 身份和生命周期管理, (3) 智能体之间的通信通道。\\n\\n## 解决方案\\n\\n```\\nTeammate lifecycle:\\n spawn -> WORKING -> IDLE -> WORKING -> ... -> SHUTDOWN\\n\\nCommunication:\\n .team/\\n config.json <- team roster + statuses\\n inbox/\\n alice.jsonl <- append-only, drain-on-read\\n bob.jsonl\\n lead.jsonl\\n\\n +--------+ send(\\\"alice\\\",\\\"bob\\\",\\\"...\\\") +--------+\\n | alice | -----------------------------> | bob |\\n | loop | bob.jsonl << {json_line} | loop |\\n +--------+ +--------+\\n ^ |\\n | BUS.read_inbox(\\\"alice\\\") |\\n +---- alice.jsonl -> read + drain ---------+\\n```\\n\\n## 工作原理\\n\\n1. TeammateManager 通过 config.json 维护团队名册。\\n\\n```python\\nclass TeammateManager:\\n def __init__(self, team_dir: Path):\\n self.dir = team_dir\\n self.dir.mkdir(exist_ok=True)\\n self.config_path = self.dir / \\\"config.json\\\"\\n self.config = self._load_config()\\n self.threads = {}\\n```\\n\\n2. `spawn()` 创建队友并在线程中启动 agent loop。\\n\\n```python\\ndef spawn(self, name: str, role: str, prompt: str) -> str:\\n member = {\\\"name\\\": name, \\\"role\\\": role, \\\"status\\\": \\\"working\\\"}\\n self.config[\\\"members\\\"].append(member)\\n self._save_config()\\n thread = threading.Thread(\\n target=self._teammate_loop,\\n args=(name, role, prompt), daemon=True)\\n thread.start()\\n return f\\\"Spawned teammate '{name}' (role: {role})\\\"\\n```\\n\\n3. MessageBus: append-only 的 JSONL 收件箱。`send()` 追加一行; `read_inbox()` 读取全部并清空。\\n\\n```python\\nclass MessageBus:\\n def send(self, sender, to, content, msg_type=\\\"message\\\", extra=None):\\n msg = {\\\"type\\\": msg_type, \\\"from\\\": sender,\\n \\\"content\\\": content, \\\"timestamp\\\": time.time()}\\n if extra:\\n msg.update(extra)\\n with open(self.dir / f\\\"{to}.jsonl\\\", \\\"a\\\") as f:\\n f.write(json.dumps(msg) + \\\"\\\\n\\\")\\n\\n def read_inbox(self, name):\\n path = self.dir / f\\\"{name}.jsonl\\\"\\n if not path.exists(): return \\\"[]\\\"\\n msgs = [json.loads(l) for l in path.read_text().strip().splitlines() if l]\\n path.write_text(\\\"\\\") # drain\\n return json.dumps(msgs, indent=2)\\n```\\n\\n4. 每个队友在每次 LLM 调用前检查收件箱, 将消息注入上下文。\\n\\n```python\\ndef _teammate_loop(self, name, role, prompt):\\n messages = [{\\\"role\\\": \\\"user\\\", \\\"content\\\": prompt}]\\n for _ in range(50):\\n inbox = BUS.read_inbox(name)\\n if inbox != \\\"[]\\\":\\n messages.append({\\\"role\\\": \\\"user\\\",\\n \\\"content\\\": f\\\"{inbox} \\\"})\\n messages.append({\\\"role\\\": \\\"assistant\\\",\\n \\\"content\\\": \\\"Noted inbox messages.\\\"})\\n response = client.messages.create(...)\\n if response.stop_reason != \\\"tool_use\\\":\\n break\\n # execute tools, append results...\\n self._find_member(name)[\\\"status\\\"] = \\\"idle\\\"\\n```\\n\\n## 相对 s08 的变更\\n\\n| 组件 | 之前 (s08) | 之后 (s09) |\\n|----------------|------------------|------------------------------------|\\n| Tools | 6 | 9 (+spawn/send/read_inbox) |\\n| 智能体数量 | 单一 | 领导 + N 个队友 |\\n| 持久化 | 无 | config.json + JSONL 收件箱 |\\n| 线程 | 后台命令 | 每线程完整 agent loop |\\n| 生命周期 | 一次性 | idle -> working -> idle |\\n| 通信 | 无 | message + broadcast |\\n\\n## 试一试\\n\\n```sh\\ncd learn-claude-code\\npython agents/s09_agent_teams.py\\n```\\n\\n试试这些 prompt (英文 prompt 对 LLM 效果更好, 也可以用中文):\\n\\n1. `Spawn alice (coder) and bob (tester). Have alice send bob a message.`\\n2. `Broadcast \\\"status update: phase 1 complete\\\" to all teammates`\\n3. `Check the lead inbox for any messages`\\n4. 输入 `/team` 查看团队名册和状态\\n5. 输入 `/inbox` 手动检查领导的收件箱\\n\"\n },\n {\n \"version\": \"s10\",\n \"locale\": \"zh\",\n \"title\": \"s10: Team Protocols (团队协议)\",\n \"content\": \"# s10: Team Protocols (团队协议)\\n\\n`s01 > s02 > s03 > s04 > s05 > s06 | s07 > s08 > s09 > [ s10 ] s11 > s12`\\n\\n> *\\\"队友之间要有统一的沟通规矩\\\"* -- 一个 request-response 模式驱动所有协商。\\n\\n## 问题\\n\\ns09 中队友能干活能通信, 但缺少结构化协调:\\n\\n**关机**: 直接杀线程会留下写了一半的文件和过期的 config.json。需要握手 -- 领导请求, 队友批准 (收尾退出) 或拒绝 (继续干)。\\n\\n**计划审批**: 领导说 \\\"重构认证模块\\\", 队友立刻开干。高风险变更应该先过审。\\n\\n两者结构一样: 一方发带唯一 ID 的请求, 另一方引用同一 ID 响应。\\n\\n## 解决方案\\n\\n```\\nShutdown Protocol Plan Approval Protocol\\n================== ======================\\n\\nLead Teammate Teammate Lead\\n | | | |\\n |--shutdown_req-->| |--plan_req------>|\\n | {req_id:\\\"abc\\\"} | | {req_id:\\\"xyz\\\"} |\\n | | | |\\n |<--shutdown_resp-| |<--plan_resp-----|\\n | {req_id:\\\"abc\\\", | | {req_id:\\\"xyz\\\", |\\n | approve:true} | | approve:true} |\\n\\nShared FSM:\\n [pending] --approve--> [approved]\\n [pending] --reject---> [rejected]\\n\\nTrackers:\\n shutdown_requests = {req_id: {target, status}}\\n plan_requests = {req_id: {from, plan, status}}\\n```\\n\\n## 工作原理\\n\\n1. 领导生成 request_id, 通过收件箱发起关机请求。\\n\\n```python\\nshutdown_requests = {}\\n\\ndef handle_shutdown_request(teammate: str) -> str:\\n req_id = str(uuid.uuid4())[:8]\\n shutdown_requests[req_id] = {\\\"target\\\": teammate, \\\"status\\\": \\\"pending\\\"}\\n BUS.send(\\\"lead\\\", teammate, \\\"Please shut down gracefully.\\\",\\n \\\"shutdown_request\\\", {\\\"request_id\\\": req_id})\\n return f\\\"Shutdown request {req_id} sent (status: pending)\\\"\\n```\\n\\n2. 队友收到请求后, 用 approve/reject 响应。\\n\\n```python\\nif tool_name == \\\"shutdown_response\\\":\\n req_id = args[\\\"request_id\\\"]\\n approve = args[\\\"approve\\\"]\\n shutdown_requests[req_id][\\\"status\\\"] = \\\"approved\\\" if approve else \\\"rejected\\\"\\n BUS.send(sender, \\\"lead\\\", args.get(\\\"reason\\\", \\\"\\\"),\\n \\\"shutdown_response\\\",\\n {\\\"request_id\\\": req_id, \\\"approve\\\": approve})\\n```\\n\\n3. 计划审批遵循完全相同的模式。队友提交计划 (生成 request_id), 领导审查 (引用同一个 request_id)。\\n\\n```python\\nplan_requests = {}\\n\\ndef handle_plan_review(request_id, approve, feedback=\\\"\\\"):\\n req = plan_requests[request_id]\\n req[\\\"status\\\"] = \\\"approved\\\" if approve else \\\"rejected\\\"\\n BUS.send(\\\"lead\\\", req[\\\"from\\\"], feedback,\\n \\\"plan_approval_response\\\",\\n {\\\"request_id\\\": request_id, \\\"approve\\\": approve})\\n```\\n\\n一个 FSM, 两种用途。同样的 `pending -> approved | rejected` 状态机可以套用到任何请求-响应协议上。\\n\\n## 相对 s09 的变更\\n\\n| 组件 | 之前 (s09) | 之后 (s10) |\\n|----------------|------------------|--------------------------------------|\\n| Tools | 9 | 12 (+shutdown_req/resp +plan) |\\n| 关机 | 仅自然退出 | 请求-响应握手 |\\n| 计划门控 | 无 | 提交/审查与审批 |\\n| 关联 | 无 | 每个请求一个 request_id |\\n| FSM | 无 | pending -> approved/rejected |\\n\\n## 试一试\\n\\n```sh\\ncd learn-claude-code\\npython agents/s10_team_protocols.py\\n```\\n\\n试试这些 prompt (英文 prompt 对 LLM 效果更好, 也可以用中文):\\n\\n1. `Spawn alice as a coder. Then request her shutdown.`\\n2. `List teammates to see alice's status after shutdown approval`\\n3. `Spawn bob with a risky refactoring task. Review and reject his plan.`\\n4. `Spawn charlie, have him submit a plan, then approve it.`\\n5. 输入 `/team` 监控状态\\n\"\n },\n {\n \"version\": \"s11\",\n \"locale\": \"zh\",\n \"title\": \"s11: Autonomous Agents (自治智能体)\",\n \"content\": \"# s11: Autonomous Agents (自治智能体)\\n\\n`s01 > s02 > s03 > s04 > s05 > s06 | s07 > s08 > s09 > s10 > [ s11 ] s12`\\n\\n> *\\\"队友自己看看板, 有活就认领\\\"* -- 不需要领导逐个分配, 自组织。\\n\\n## 问题\\n\\ns09-s10 中, 队友只在被明确指派时才动。领导得给每个队友写 prompt, 任务看板上 10 个未认领的任务得手动分配。这扩展不了。\\n\\n真正的自治: 队友自己扫描任务看板, 认领没人做的任务, 做完再找下一个。\\n\\n一个细节: 上下文压缩 (s06) 后智能体可能忘了自己是谁。身份重注入解决这个问题。\\n\\n## 解决方案\\n\\n```\\nTeammate lifecycle with idle cycle:\\n\\n+-------+\\n| spawn |\\n+---+---+\\n |\\n v\\n+-------+ tool_use +-------+\\n| WORK | <------------- | LLM |\\n+---+---+ +-------+\\n |\\n | stop_reason != tool_use (or idle tool called)\\n v\\n+--------+\\n| IDLE | poll every 5s for up to 60s\\n+---+----+\\n |\\n +---> check inbox --> message? ----------> WORK\\n |\\n +---> scan .tasks/ --> unclaimed? -------> claim -> WORK\\n |\\n +---> 60s timeout ----------------------> SHUTDOWN\\n\\nIdentity re-injection after compression:\\n if len(messages) <= 3:\\n messages.insert(0, identity_block)\\n```\\n\\n## 工作原理\\n\\n1. 队友循环分两个阶段: WORK 和 IDLE。LLM 停止调用工具 (或调用了 `idle`) 时, 进入 IDLE。\\n\\n```python\\ndef _loop(self, name, role, prompt):\\n while True:\\n # -- WORK PHASE --\\n messages = [{\\\"role\\\": \\\"user\\\", \\\"content\\\": prompt}]\\n for _ in range(50):\\n response = client.messages.create(...)\\n if response.stop_reason != \\\"tool_use\\\":\\n break\\n # execute tools...\\n if idle_requested:\\n break\\n\\n # -- IDLE PHASE --\\n self._set_status(name, \\\"idle\\\")\\n resume = self._idle_poll(name, messages)\\n if not resume:\\n self._set_status(name, \\\"shutdown\\\")\\n return\\n self._set_status(name, \\\"working\\\")\\n```\\n\\n2. 空闲阶段循环轮询收件箱和任务看板。\\n\\n```python\\ndef _idle_poll(self, name, messages):\\n for _ in range(IDLE_TIMEOUT // POLL_INTERVAL): # 60s / 5s = 12\\n time.sleep(POLL_INTERVAL)\\n inbox = BUS.read_inbox(name)\\n if inbox:\\n messages.append({\\\"role\\\": \\\"user\\\",\\n \\\"content\\\": f\\\"{inbox} \\\"})\\n return True\\n unclaimed = scan_unclaimed_tasks()\\n if unclaimed:\\n claim_task(unclaimed[0][\\\"id\\\"], name)\\n messages.append({\\\"role\\\": \\\"user\\\",\\n \\\"content\\\": f\\\"Task #{unclaimed[0]['id']}: \\\"\\n f\\\"{unclaimed[0]['subject']} \\\"})\\n return True\\n return False # timeout -> shutdown\\n```\\n\\n3. 任务看板扫描: 找 pending 状态、无 owner、未被阻塞的任务。\\n\\n```python\\ndef scan_unclaimed_tasks() -> list:\\n unclaimed = []\\n for f in sorted(TASKS_DIR.glob(\\\"task_*.json\\\")):\\n task = json.loads(f.read_text())\\n if (task.get(\\\"status\\\") == \\\"pending\\\"\\n and not task.get(\\\"owner\\\")\\n and not task.get(\\\"blockedBy\\\")):\\n unclaimed.append(task)\\n return unclaimed\\n```\\n\\n4. 身份重注入: 上下文过短 (说明发生了压缩) 时, 在开头插入身份块。\\n\\n```python\\nif len(messages) <= 3:\\n messages.insert(0, {\\\"role\\\": \\\"user\\\",\\n \\\"content\\\": f\\\"You are '{name}', role: {role}, \\\"\\n f\\\"team: {team_name}. Continue your work. \\\"})\\n messages.insert(1, {\\\"role\\\": \\\"assistant\\\",\\n \\\"content\\\": f\\\"I am {name}. Continuing.\\\"})\\n```\\n\\n## 相对 s10 的变更\\n\\n| 组件 | 之前 (s10) | 之后 (s11) |\\n|----------------|------------------|----------------------------------|\\n| Tools | 12 | 14 (+idle, +claim_task) |\\n| 自治性 | 领导指派 | 自组织 |\\n| 空闲阶段 | 无 | 轮询收件箱 + 任务看板 |\\n| 任务认领 | 仅手动 | 自动认领未分配任务 |\\n| 身份 | 系统提示 | + 压缩后重注入 |\\n| 超时 | 无 | 60 秒空闲 -> 自动关机 |\\n\\n## 试一试\\n\\n```sh\\ncd learn-claude-code\\npython agents/s11_autonomous_agents.py\\n```\\n\\n试试这些 prompt (英文 prompt 对 LLM 效果更好, 也可以用中文):\\n\\n1. `Create 3 tasks on the board, then spawn alice and bob. Watch them auto-claim.`\\n2. `Spawn a coder teammate and let it find work from the task board itself`\\n3. `Create tasks with dependencies. Watch teammates respect the blocked order.`\\n4. 输入 `/tasks` 查看带 owner 的任务看板\\n5. 输入 `/team` 监控谁在工作、谁在空闲\\n\"\n },\n {\n \"version\": \"s12\",\n \"locale\": \"zh\",\n \"title\": \"s12: Worktree + Task Isolation (Worktree 任务隔离)\",\n \"content\": \"# s12: Worktree + Task Isolation (Worktree 任务隔离)\\n\\n`s01 > s02 > s03 > s04 > s05 > s06 | s07 > s08 > s09 > s10 > s11 > [ s12 ]`\\n\\n> *\\\"各干各的目录, 互不干扰\\\"* -- 任务管目标, worktree 管目录, 按 ID 绑定。\\n\\n## 问题\\n\\n到 s11, 智能体已经能自主认领和完成任务。但所有任务共享一个目录。两个智能体同时重构不同模块 -- A 改 `config.py`, B 也改 `config.py`, 未提交的改动互相污染, 谁也没法干净回滚。\\n\\n任务板管 \\\"做什么\\\" 但不管 \\\"在哪做\\\"。解法: 给每个任务一个独立的 git worktree 目录, 用任务 ID 把两边关联起来。\\n\\n## 解决方案\\n\\n```\\nControl plane (.tasks/) Execution plane (.worktrees/)\\n+------------------+ +------------------------+\\n| task_1.json | | auth-refactor/ |\\n| status: in_progress <------> branch: wt/auth-refactor\\n| worktree: \\\"auth-refactor\\\" | task_id: 1 |\\n+------------------+ +------------------------+\\n| task_2.json | | ui-login/ |\\n| status: pending <------> branch: wt/ui-login\\n| worktree: \\\"ui-login\\\" | task_id: 2 |\\n+------------------+ +------------------------+\\n |\\n index.json (worktree registry)\\n events.jsonl (lifecycle log)\\n\\nState machines:\\n Task: pending -> in_progress -> completed\\n Worktree: absent -> active -> removed | kept\\n```\\n\\n## 工作原理\\n\\n1. **创建任务。** 先把目标持久化。\\n\\n```python\\nTASKS.create(\\\"Implement auth refactor\\\")\\n# -> .tasks/task_1.json status=pending worktree=\\\"\\\"\\n```\\n\\n2. **创建 worktree 并绑定任务。** 传入 `task_id` 自动将任务推进到 `in_progress`。\\n\\n```python\\nWORKTREES.create(\\\"auth-refactor\\\", task_id=1)\\n# -> git worktree add -b wt/auth-refactor .worktrees/auth-refactor HEAD\\n# -> index.json gets new entry, task_1.json gets worktree=\\\"auth-refactor\\\"\\n```\\n\\n绑定同时写入两侧状态:\\n\\n```python\\ndef bind_worktree(self, task_id, worktree):\\n task = self._load(task_id)\\n task[\\\"worktree\\\"] = worktree\\n if task[\\\"status\\\"] == \\\"pending\\\":\\n task[\\\"status\\\"] = \\\"in_progress\\\"\\n self._save(task)\\n```\\n\\n3. **在 worktree 中执行命令。** `cwd` 指向隔离目录。\\n\\n```python\\nsubprocess.run(command, shell=True, cwd=worktree_path,\\n capture_output=True, text=True, timeout=300)\\n```\\n\\n4. **收尾。** 两种选择:\\n - `worktree_keep(name)` -- 保留目录供后续使用。\\n - `worktree_remove(name, complete_task=True)` -- 删除目录, 完成绑定任务, 发出事件。一个调用搞定拆除 + 完成。\\n\\n```python\\ndef remove(self, name, force=False, complete_task=False):\\n self._run_git([\\\"worktree\\\", \\\"remove\\\", wt[\\\"path\\\"]])\\n if complete_task and wt.get(\\\"task_id\\\") is not None:\\n self.tasks.update(wt[\\\"task_id\\\"], status=\\\"completed\\\")\\n self.tasks.unbind_worktree(wt[\\\"task_id\\\"])\\n self.events.emit(\\\"task.completed\\\", ...)\\n```\\n\\n5. **事件流。** 每个生命周期步骤写入 `.worktrees/events.jsonl`:\\n\\n```json\\n{\\n \\\"event\\\": \\\"worktree.remove.after\\\",\\n \\\"task\\\": {\\\"id\\\": 1, \\\"status\\\": \\\"completed\\\"},\\n \\\"worktree\\\": {\\\"name\\\": \\\"auth-refactor\\\", \\\"status\\\": \\\"removed\\\"},\\n \\\"ts\\\": 1730000000\\n}\\n```\\n\\n事件类型: `worktree.create.before/after/failed`, `worktree.remove.before/after/failed`, `worktree.keep`, `task.completed`。\\n\\n崩溃后从 `.tasks/` + `.worktrees/index.json` 重建现场。会话记忆是易失的; 磁盘状态是持久的。\\n\\n## 相对 s11 的变更\\n\\n| 组件 | 之前 (s11) | 之后 (s12) |\\n|--------------------|----------------------------|----------------------------------------------|\\n| 协调 | 任务板 (owner/status) | 任务板 + worktree 显式绑定 |\\n| 执行范围 | 共享目录 | 每个任务独立目录 |\\n| 可恢复性 | 仅任务状态 | 任务状态 + worktree 索引 |\\n| 收尾 | 任务完成 | 任务完成 + 显式 keep/remove |\\n| 生命周期可见性 | 隐式日志 | `.worktrees/events.jsonl` 显式事件流 |\\n\\n## 试一试\\n\\n```sh\\ncd learn-claude-code\\npython agents/s12_worktree_task_isolation.py\\n```\\n\\n试试这些 prompt (英文 prompt 对 LLM 效果更好, 也可以用中文):\\n\\n1. `Create tasks for backend auth and frontend login page, then list tasks.`\\n2. `Create worktree \\\"auth-refactor\\\" for task 1, then bind task 2 to a new worktree \\\"ui-login\\\".`\\n3. `Run \\\"git status --short\\\" in worktree \\\"auth-refactor\\\".`\\n4. `Keep worktree \\\"ui-login\\\", then list worktrees and inspect events.`\\n5. `Remove worktree \\\"auth-refactor\\\" with complete_task=true, then list tasks/worktrees/events.`\\n\"\n },\n {\n \"version\": \"s01\",\n \"locale\": \"ja\",\n \"title\": \"s01: The Agent Loop\",\n \"content\": \"# s01: The Agent Loop\\n\\n`[ s01 ] s02 > s03 > s04 > s05 > s06 | s07 > s08 > s09 > s10 > s11 > s12`\\n\\n> *\\\"One loop & Bash is all you need\\\"* -- 1つのツール + 1つのループ = エージェント。\\n\\n## 問題\\n\\n言語モデルはコードについて推論できるが、現実世界に触れられない。ファイルを読めず、テストを実行できず、エラーを確認できない。ループがなければ、ツール呼び出しのたびにユーザーが手動で結果をコピーペーストする必要がある。つまりユーザー自身がループになる。\\n\\n## 解決策\\n\\n```\\n+--------+ +-------+ +---------+\\n| User | ---> | LLM | ---> | Tool |\\n| prompt | | | | execute |\\n+--------+ +---+---+ +----+----+\\n ^ |\\n | tool_result |\\n +----------------+\\n (loop until stop_reason != \\\"tool_use\\\")\\n```\\n\\n1つの終了条件がフロー全体を制御する。モデルがツール呼び出しを止めるまでループが回り続ける。\\n\\n## 仕組み\\n\\n1. ユーザーのプロンプトが最初のメッセージになる。\\n\\n```python\\nmessages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": query})\\n```\\n\\n2. メッセージとツール定義をLLMに送信する。\\n\\n```python\\nresponse = client.messages.create(\\n model=MODEL, system=SYSTEM, messages=messages,\\n tools=TOOLS, max_tokens=8000,\\n)\\n```\\n\\n3. アシスタントのレスポンスを追加し、`stop_reason`を確認する。ツールが呼ばれなければ終了。\\n\\n```python\\nmessages.append({\\\"role\\\": \\\"assistant\\\", \\\"content\\\": response.content})\\nif response.stop_reason != \\\"tool_use\\\":\\n return\\n```\\n\\n4. 各ツール呼び出しを実行し、結果を収集してuserメッセージとして追加。ステップ2に戻る。\\n\\n```python\\nresults = []\\nfor block in response.content:\\n if block.type == \\\"tool_use\\\":\\n output = run_bash(block.input[\\\"command\\\"])\\n results.append({\\n \\\"type\\\": \\\"tool_result\\\",\\n \\\"tool_use_id\\\": block.id,\\n \\\"content\\\": output,\\n })\\nmessages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": results})\\n```\\n\\n1つの関数にまとめると:\\n\\n```python\\ndef agent_loop(query):\\n messages = [{\\\"role\\\": \\\"user\\\", \\\"content\\\": query}]\\n while True:\\n response = client.messages.create(\\n model=MODEL, system=SYSTEM, messages=messages,\\n tools=TOOLS, max_tokens=8000,\\n )\\n messages.append({\\\"role\\\": \\\"assistant\\\", \\\"content\\\": response.content})\\n\\n if response.stop_reason != \\\"tool_use\\\":\\n return\\n\\n results = []\\n for block in response.content:\\n if block.type == \\\"tool_use\\\":\\n output = run_bash(block.input[\\\"command\\\"])\\n results.append({\\n \\\"type\\\": \\\"tool_result\\\",\\n \\\"tool_use_id\\\": block.id,\\n \\\"content\\\": output,\\n })\\n messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": results})\\n```\\n\\nこれでエージェント全体が30行未満に収まる。本コースの残りはすべてこのループの上に積み重なる -- ループ自体は変わらない。\\n\\n## 変更点\\n\\n| Component | Before | After |\\n|---------------|------------|--------------------------------|\\n| Agent loop | (none) | `while True` + stop_reason |\\n| Tools | (none) | `bash` (one tool) |\\n| Messages | (none) | Accumulating list |\\n| Control flow | (none) | `stop_reason != \\\"tool_use\\\"` |\\n\\n## 試してみる\\n\\n```sh\\ncd learn-claude-code\\npython agents/s01_agent_loop.py\\n```\\n\\n1. `Create a file called hello.py that prints \\\"Hello, World!\\\"`\\n2. `List all Python files in this directory`\\n3. `What is the current git branch?`\\n4. `Create a directory called test_output and write 3 files in it`\\n\"\n },\n {\n \"version\": \"s02\",\n \"locale\": \"ja\",\n \"title\": \"s02: Tool Use\",\n \"content\": \"# s02: Tool Use\\n\\n`s01 > [ s02 ] s03 > s04 > s05 > s06 | s07 > s08 > s09 > s10 > s11 > s12`\\n\\n> *\\\"ツールを足すなら、ハンドラーを1つ足すだけ\\\"* -- ループは変わらない。新ツールは dispatch map に登録するだけ。\\n\\n## 問題\\n\\n`bash`だけでは、エージェントは何でもシェル経由で行う。`cat`は予測不能に切り詰め、`sed`は特殊文字で壊れ、すべてのbash呼び出しが制約のないセキュリティ面になる。`read_file`や`write_file`のような専用ツールなら、ツールレベルでパスのサンドボックス化を強制できる。\\n\\n重要な点: ツールを追加してもループの変更は不要。\\n\\n## 解決策\\n\\n```\\n+--------+ +-------+ +------------------+\\n| User | ---> | LLM | ---> | Tool Dispatch |\\n| prompt | | | | { |\\n+--------+ +---+---+ | bash: run_bash |\\n ^ | read: run_read |\\n | | write: run_wr |\\n +-----------+ edit: run_edit |\\n tool_result | } |\\n +------------------+\\n\\nThe dispatch map is a dict: {tool_name: handler_function}.\\nOne lookup replaces any if/elif chain.\\n```\\n\\n## 仕組み\\n\\n1. 各ツールにハンドラ関数を定義する。パスのサンドボックス化でワークスペース外への脱出を防ぐ。\\n\\n```python\\ndef safe_path(p: str) -> Path:\\n path = (WORKDIR / p).resolve()\\n if not path.is_relative_to(WORKDIR):\\n raise ValueError(f\\\"Path escapes workspace: {p}\\\")\\n return path\\n\\ndef run_read(path: str, limit: int = None) -> str:\\n text = safe_path(path).read_text()\\n lines = text.splitlines()\\n if limit and limit < len(lines):\\n lines = lines[:limit]\\n return \\\"\\\\n\\\".join(lines)[:50000]\\n```\\n\\n2. ディスパッチマップがツール名とハンドラを結びつける。\\n\\n```python\\nTOOL_HANDLERS = {\\n \\\"bash\\\": lambda **kw: run_bash(kw[\\\"command\\\"]),\\n \\\"read_file\\\": lambda **kw: run_read(kw[\\\"path\\\"], kw.get(\\\"limit\\\")),\\n \\\"write_file\\\": lambda **kw: run_write(kw[\\\"path\\\"], kw[\\\"content\\\"]),\\n \\\"edit_file\\\": lambda **kw: run_edit(kw[\\\"path\\\"], kw[\\\"old_text\\\"],\\n kw[\\\"new_text\\\"]),\\n}\\n```\\n\\n3. ループ内で名前によりハンドラをルックアップする。ループ本体はs01から不変。\\n\\n```python\\nfor block in response.content:\\n if block.type == \\\"tool_use\\\":\\n handler = TOOL_HANDLERS.get(block.name)\\n output = handler(**block.input) if handler \\\\\\n else f\\\"Unknown tool: {block.name}\\\"\\n results.append({\\n \\\"type\\\": \\\"tool_result\\\",\\n \\\"tool_use_id\\\": block.id,\\n \\\"content\\\": output,\\n })\\n```\\n\\nツール追加 = ハンドラ追加 + スキーマ追加。ループは決して変わらない。\\n\\n## s01からの変更点\\n\\n| Component | Before (s01) | After (s02) |\\n|----------------|--------------------|----------------------------|\\n| Tools | 1 (bash only) | 4 (bash, read, write, edit)|\\n| Dispatch | Hardcoded bash call | `TOOL_HANDLERS` dict |\\n| Path safety | None | `safe_path()` sandbox |\\n| Agent loop | Unchanged | Unchanged |\\n\\n## 試してみる\\n\\n```sh\\ncd learn-claude-code\\npython agents/s02_tool_use.py\\n```\\n\\n1. `Read the file requirements.txt`\\n2. `Create a file called greet.py with a greet(name) function`\\n3. `Edit greet.py to add a docstring to the function`\\n4. `Read greet.py to verify the edit worked`\\n\"\n },\n {\n \"version\": \"s03\",\n \"locale\": \"ja\",\n \"title\": \"s03: TodoWrite\",\n \"content\": \"# s03: TodoWrite\\n\\n`s01 > s02 > [ s03 ] s04 > s05 > s06 | s07 > s08 > s09 > s10 > s11 > s12`\\n\\n> *\\\"計画のないエージェントは行き当たりばったり\\\"* -- まずステップを書き出し、それから実行。\\n\\n## 問題\\n\\nマルチステップのタスクで、モデルは途中で迷子になる。作業を繰り返したり、ステップを飛ばしたり、脱線したりする。長い会話になるほど悪化する -- ツール結果がコンテキストを埋めるにつれ、システムプロンプトの影響力が薄れる。10ステップのリファクタリングでステップ1-3を完了した後、残りを忘れて即興を始めてしまう。\\n\\n## 解決策\\n\\n```\\n+--------+ +-------+ +---------+\\n| User | ---> | LLM | ---> | Tools |\\n| prompt | | | | + todo |\\n+--------+ +---+---+ +----+----+\\n ^ |\\n | tool_result |\\n +----------------+\\n |\\n +-----------+-----------+\\n | TodoManager state |\\n | [ ] task A |\\n | [>] task B <- doing |\\n | [x] task C |\\n +-----------------------+\\n |\\n if rounds_since_todo >= 3:\\n inject into tool_result\\n```\\n\\n## 仕組み\\n\\n1. TodoManagerはアイテムのリストをステータス付きで保持する。`in_progress`にできるのは同時に1つだけ。\\n\\n```python\\nclass TodoManager:\\n def update(self, items: list) -> str:\\n validated, in_progress_count = [], 0\\n for item in items:\\n status = item.get(\\\"status\\\", \\\"pending\\\")\\n if status == \\\"in_progress\\\":\\n in_progress_count += 1\\n validated.append({\\\"id\\\": item[\\\"id\\\"], \\\"text\\\": item[\\\"text\\\"],\\n \\\"status\\\": status})\\n if in_progress_count > 1:\\n raise ValueError(\\\"Only one task can be in_progress\\\")\\n self.items = validated\\n return self.render()\\n```\\n\\n2. `todo`ツールは他のツールと同様にディスパッチマップに追加される。\\n\\n```python\\nTOOL_HANDLERS = {\\n # ...base tools...\\n \\\"todo\\\": lambda **kw: TODO.update(kw[\\\"items\\\"]),\\n}\\n```\\n\\n3. nagリマインダーが、モデルが3ラウンド以上`todo`を呼ばなかった場合にナッジを注入する。\\n\\n```python\\nif rounds_since_todo >= 3 and messages:\\n last = messages[-1]\\n if last[\\\"role\\\"] == \\\"user\\\" and isinstance(last.get(\\\"content\\\"), list):\\n last[\\\"content\\\"].insert(0, {\\n \\\"type\\\": \\\"text\\\",\\n \\\"text\\\": \\\"Update your todos. \\\",\\n })\\n```\\n\\n「一度にin_progressは1つだけ」の制約が逐次的な集中を強制し、nagリマインダーが説明責任を生む。\\n\\n## s02からの変更点\\n\\n| Component | Before (s02) | After (s03) |\\n|----------------|------------------|----------------------------|\\n| Tools | 4 | 5 (+todo) |\\n| Planning | None | TodoManager with statuses |\\n| Nag injection | None | `` after 3 rounds|\\n| Agent loop | Simple dispatch | + rounds_since_todo counter|\\n\\n## 試してみる\\n\\n```sh\\ncd learn-claude-code\\npython agents/s03_todo_write.py\\n```\\n\\n1. `Refactor the file hello.py: add type hints, docstrings, and a main guard`\\n2. `Create a Python package with __init__.py, utils.py, and tests/test_utils.py`\\n3. `Review all Python files and fix any style issues`\\n\"\n },\n {\n \"version\": \"s04\",\n \"locale\": \"ja\",\n \"title\": \"s04: Subagents\",\n \"content\": \"# s04: Subagents\\n\\n`s01 > s02 > s03 > [ s04 ] s05 > s06 | s07 > s08 > s09 > s10 > s11 > s12`\\n\\n> *\\\"大きなタスクを分割し、各サブタスクにクリーンなコンテキストを\\\"* -- サブエージェントは独立した messages[] を使い、メイン会話を汚さない。\\n\\n## 問題\\n\\nエージェントが作業するにつれ、messages配列は膨張し続ける。すべてのファイル読み取り、すべてのbash出力がコンテキストに永久に残る。「このプロジェクトはどのテストフレームワークを使っているか」という質問は5つのファイルを読む必要があるかもしれないが、親に必要なのは「pytest」という答えだけだ。\\n\\n## 解決策\\n\\n```\\nParent agent Subagent\\n+------------------+ +------------------+\\n| messages=[...] | | messages=[] | <-- fresh\\n| | dispatch | |\\n| tool: task | ----------> | while tool_use: |\\n| prompt=\\\"...\\\" | | call tools |\\n| | summary | append results |\\n| result = \\\"...\\\" | <---------- | return last text |\\n+------------------+ +------------------+\\n\\nParent context stays clean. Subagent context is discarded.\\n```\\n\\n## 仕組み\\n\\n1. 親に`task`ツールを追加する。子は`task`を除くすべての基本ツールを取得する(再帰的な生成は不可)。\\n\\n```python\\nPARENT_TOOLS = CHILD_TOOLS + [\\n {\\\"name\\\": \\\"task\\\",\\n \\\"description\\\": \\\"Spawn a subagent with fresh context.\\\",\\n \\\"input_schema\\\": {\\n \\\"type\\\": \\\"object\\\",\\n \\\"properties\\\": {\\\"prompt\\\": {\\\"type\\\": \\\"string\\\"}},\\n \\\"required\\\": [\\\"prompt\\\"],\\n }},\\n]\\n```\\n\\n2. サブエージェントは`messages=[]`で開始し、自身のループを実行する。最終テキストだけが親に返る。\\n\\n```python\\ndef run_subagent(prompt: str) -> str:\\n sub_messages = [{\\\"role\\\": \\\"user\\\", \\\"content\\\": prompt}]\\n for _ in range(30): # safety limit\\n response = client.messages.create(\\n model=MODEL, system=SUBAGENT_SYSTEM,\\n messages=sub_messages,\\n tools=CHILD_TOOLS, max_tokens=8000,\\n )\\n sub_messages.append({\\\"role\\\": \\\"assistant\\\",\\n \\\"content\\\": response.content})\\n if response.stop_reason != \\\"tool_use\\\":\\n break\\n results = []\\n for block in response.content:\\n if block.type == \\\"tool_use\\\":\\n handler = TOOL_HANDLERS.get(block.name)\\n output = handler(**block.input)\\n results.append({\\\"type\\\": \\\"tool_result\\\",\\n \\\"tool_use_id\\\": block.id,\\n \\\"content\\\": str(output)[:50000]})\\n sub_messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": results})\\n return \\\"\\\".join(\\n b.text for b in response.content if hasattr(b, \\\"text\\\")\\n ) or \\\"(no summary)\\\"\\n```\\n\\n子のメッセージ履歴全体(30回以上のツール呼び出し)は破棄される。親は1段落の要約を通常の`tool_result`として受け取る。\\n\\n## s03からの変更点\\n\\n| Component | Before (s03) | After (s04) |\\n|----------------|------------------|---------------------------|\\n| Tools | 5 | 5 (base) + task (parent) |\\n| Context | Single shared | Parent + child isolation |\\n| Subagent | None | `run_subagent()` function |\\n| Return value | N/A | Summary text only |\\n\\n## 試してみる\\n\\n```sh\\ncd learn-claude-code\\npython agents/s04_subagent.py\\n```\\n\\n1. `Use a subtask to find what testing framework this project uses`\\n2. `Delegate: read all .py files and summarize what each one does`\\n3. `Use a task to create a new module, then verify it from here`\\n\"\n },\n {\n \"version\": \"s05\",\n \"locale\": \"ja\",\n \"title\": \"s05: Skills\",\n \"content\": \"# s05: Skills\\n\\n`s01 > s02 > s03 > s04 > [ s05 ] s06 | s07 > s08 > s09 > s10 > s11 > s12`\\n\\n> *\\\"必要な知識を、必要な時に読み込む\\\"* -- system prompt ではなく tool_result で注入。\\n\\n## 問題\\n\\nエージェントにドメイン固有のワークフローを遵守させたい: gitの規約、テストパターン、コードレビューチェックリスト。すべてをシステムプロンプトに入れると、使われないスキルにトークンを浪費する。10スキル x 2000トークン = 20,000トークン、ほとんどが任意のタスクに無関係だ。\\n\\n## 解決策\\n\\n```\\nSystem prompt (Layer 1 -- always present):\\n+--------------------------------------+\\n| You are a coding agent. |\\n| Skills available: |\\n| - git: Git workflow helpers | ~100 tokens/skill\\n| - test: Testing best practices |\\n+--------------------------------------+\\n\\nWhen model calls load_skill(\\\"git\\\"):\\n+--------------------------------------+\\n| tool_result (Layer 2 -- on demand): |\\n| |\\n| Full git workflow instructions... | ~2000 tokens\\n| Step 1: ... |\\n| |\\n+--------------------------------------+\\n```\\n\\n第1層: スキル*名*をシステムプロンプトに(低コスト)。第2層: スキル*本体*をtool_resultに(オンデマンド)。\\n\\n## 仕組み\\n\\n1. 各スキルは `SKILL.md` ファイルを含むディレクトリとして配置される。\\n\\n```\\nskills/\\n pdf/\\n SKILL.md # ---\\\\n name: pdf\\\\n description: Process PDF files\\\\n ---\\\\n ...\\n code-review/\\n SKILL.md # ---\\\\n name: code-review\\\\n description: Review code\\\\n ---\\\\n ...\\n```\\n\\n2. SkillLoaderが `SKILL.md` を再帰的に探索し、ディレクトリ名をスキル識別子として使用する。\\n\\n```python\\nclass SkillLoader:\\n def __init__(self, skills_dir: Path):\\n self.skills = {}\\n for f in sorted(skills_dir.rglob(\\\"SKILL.md\\\")):\\n text = f.read_text()\\n meta, body = self._parse_frontmatter(text)\\n name = meta.get(\\\"name\\\", f.parent.name)\\n self.skills[name] = {\\\"meta\\\": meta, \\\"body\\\": body}\\n\\n def get_descriptions(self) -> str:\\n lines = []\\n for name, skill in self.skills.items():\\n desc = skill[\\\"meta\\\"].get(\\\"description\\\", \\\"\\\")\\n lines.append(f\\\" - {name}: {desc}\\\")\\n return \\\"\\\\n\\\".join(lines)\\n\\n def get_content(self, name: str) -> str:\\n skill = self.skills.get(name)\\n if not skill:\\n return f\\\"Error: Unknown skill '{name}'.\\\"\\n return f\\\"\\\\n{skill['body']}\\\\n \\\"\\n```\\n\\n3. 第1層はシステムプロンプトに配置。第2層は通常のツールハンドラ。\\n\\n```python\\nSYSTEM = f\\\"\\\"\\\"You are a coding agent at {WORKDIR}.\\nSkills available:\\n{SKILL_LOADER.get_descriptions()}\\\"\\\"\\\"\\n\\nTOOL_HANDLERS = {\\n # ...base tools...\\n \\\"load_skill\\\": lambda **kw: SKILL_LOADER.get_content(kw[\\\"name\\\"]),\\n}\\n```\\n\\nモデルはどのスキルが存在するかを知り(低コスト)、関連する時にだけ読み込む(高コスト)。\\n\\n## s04からの変更点\\n\\n| Component | Before (s04) | After (s05) |\\n|----------------|------------------|----------------------------|\\n| Tools | 5 (base + task) | 5 (base + load_skill) |\\n| System prompt | Static string | + skill descriptions |\\n| Knowledge | None | skills/\\\\*/SKILL.md files |\\n| Injection | None | Two-layer (system + result)|\\n\\n## 試してみる\\n\\n```sh\\ncd learn-claude-code\\npython agents/s05_skill_loading.py\\n```\\n\\n1. `What skills are available?`\\n2. `Load the agent-builder skill and follow its instructions`\\n3. `I need to do a code review -- load the relevant skill first`\\n4. `Build an MCP server using the mcp-builder skill`\\n\"\n },\n {\n \"version\": \"s06\",\n \"locale\": \"ja\",\n \"title\": \"s06: Context Compact\",\n \"content\": \"# s06: Context Compact\\n\\n`s01 > s02 > s03 > s04 > s05 > [ s06 ] | s07 > s08 > s09 > s10 > s11 > s12`\\n\\n> *\\\"コンテキストはいつか溢れる、空ける手段が要る\\\"* -- 3層圧縮で無限セッションを実現。\\n\\n## 問題\\n\\nコンテキストウィンドウは有限だ。1000行のファイルに対する`read_file`1回で約4000トークンを消費する。30ファイルを読み20回のbashコマンドを実行すると、100,000トークン超。圧縮なしでは、エージェントは大規模コードベースで作業できない。\\n\\n## 解決策\\n\\n積極性を段階的に上げる3層構成:\\n\\n```\\nEvery turn:\\n+------------------+\\n| Tool call result |\\n+------------------+\\n |\\n v\\n[Layer 1: micro_compact] (silent, every turn)\\n Replace tool_result > 3 turns old\\n with \\\"[Previous: used {tool_name}]\\\"\\n |\\n v\\n[Check: tokens > 50000?]\\n | |\\n no yes\\n | |\\n v v\\ncontinue [Layer 2: auto_compact]\\n Save transcript to .transcripts/\\n LLM summarizes conversation.\\n Replace all messages with [summary].\\n |\\n v\\n [Layer 3: compact tool]\\n Model calls compact explicitly.\\n Same summarization as auto_compact.\\n```\\n\\n## 仕組み\\n\\n1. **第1層 -- micro_compact**: 各LLM呼び出しの前に、古いツール結果をプレースホルダーに置換する。\\n\\n```python\\ndef micro_compact(messages: list) -> list:\\n tool_results = []\\n for i, msg in enumerate(messages):\\n if msg[\\\"role\\\"] == \\\"user\\\" and isinstance(msg.get(\\\"content\\\"), list):\\n for j, part in enumerate(msg[\\\"content\\\"]):\\n if isinstance(part, dict) and part.get(\\\"type\\\") == \\\"tool_result\\\":\\n tool_results.append((i, j, part))\\n if len(tool_results) <= KEEP_RECENT:\\n return messages\\n for _, _, part in tool_results[:-KEEP_RECENT]:\\n if len(part.get(\\\"content\\\", \\\"\\\")) > 100:\\n part[\\\"content\\\"] = f\\\"[Previous: used {tool_name}]\\\"\\n return messages\\n```\\n\\n2. **第2層 -- auto_compact**: トークンが閾値を超えたら、完全なトランスクリプトをディスクに保存し、LLMに要約を依頼する。\\n\\n```python\\ndef auto_compact(messages: list) -> list:\\n # Save transcript for recovery\\n transcript_path = TRANSCRIPT_DIR / f\\\"transcript_{int(time.time())}.jsonl\\\"\\n with open(transcript_path, \\\"w\\\") as f:\\n for msg in messages:\\n f.write(json.dumps(msg, default=str) + \\\"\\\\n\\\")\\n # LLM summarizes\\n response = client.messages.create(\\n model=MODEL,\\n messages=[{\\\"role\\\": \\\"user\\\", \\\"content\\\":\\n \\\"Summarize this conversation for continuity...\\\"\\n + json.dumps(messages, default=str)[:80000]}],\\n max_tokens=2000,\\n )\\n return [\\n {\\\"role\\\": \\\"user\\\", \\\"content\\\": f\\\"[Compressed]\\\\n\\\\n{response.content[0].text}\\\"},\\n {\\\"role\\\": \\\"assistant\\\", \\\"content\\\": \\\"Understood. Continuing.\\\"},\\n ]\\n```\\n\\n3. **第3層 -- manual compact**: `compact`ツールが同じ要約処理をオンデマンドでトリガーする。\\n\\n4. ループが3層すべてを統合する:\\n\\n```python\\ndef agent_loop(messages: list):\\n while True:\\n micro_compact(messages) # Layer 1\\n if estimate_tokens(messages) > THRESHOLD:\\n messages[:] = auto_compact(messages) # Layer 2\\n response = client.messages.create(...)\\n # ... tool execution ...\\n if manual_compact:\\n messages[:] = auto_compact(messages) # Layer 3\\n```\\n\\nトランスクリプトがディスク上に完全な履歴を保持する。何も真に失われず、アクティブなコンテキストの外に移動されるだけ。\\n\\n## s05からの変更点\\n\\n| Component | Before (s05) | After (s06) |\\n|----------------|------------------|----------------------------|\\n| Tools | 5 | 5 (base + compact) |\\n| Context mgmt | None | Three-layer compression |\\n| Micro-compact | None | Old results -> placeholders|\\n| Auto-compact | None | Token threshold trigger |\\n| Transcripts | None | Saved to .transcripts/ |\\n\\n## 試してみる\\n\\n```sh\\ncd learn-claude-code\\npython agents/s06_context_compact.py\\n```\\n\\n1. `Read every Python file in the agents/ directory one by one` (micro-compactが古い結果を置換するのを観察する)\\n2. `Keep reading files until compression triggers automatically`\\n3. `Use the compact tool to manually compress the conversation`\\n\"\n },\n {\n \"version\": \"s07\",\n \"locale\": \"ja\",\n \"title\": \"s07: Task System\",\n \"content\": \"# s07: Task System\\n\\n`s01 > s02 > s03 > s04 > s05 > s06 | [ s07 ] s08 > s09 > s10 > s11 > s12`\\n\\n> *\\\"大きな目標を小タスクに分解し、順序付けし、ディスクに記録する\\\"* -- ファイルベースのタスクグラフ、マルチエージェント協調の基盤。\\n\\n## 問題\\n\\ns03のTodoManagerはメモリ上のフラットなチェックリストに過ぎない: 順序なし、依存関係なし、ステータスは完了か未完了のみ。実際の目標には構造がある -- タスクBはタスクAに依存し、タスクCとDは並行実行でき、タスクEはCとDの両方を待つ。\\n\\n明示的な関係がなければ、エージェントは何が実行可能で、何がブロックされ、何が同時に走れるかを判断できない。しかもリストはメモリ上にしかないため、コンテキスト圧縮(s06)で消える。\\n\\n## 解決策\\n\\nフラットなチェックリストをディスクに永続化する**タスクグラフ**に昇格させる。各タスクは1つのJSONファイルで、ステータス・前方依存(`blockedBy`)・後方依存(`blocks`)を持つ。タスクグラフは常に3つの問いに答える:\\n\\n- **何が実行可能か?** -- `pending`ステータスで`blockedBy`が空のタスク。\\n- **何がブロックされているか?** -- 未完了の依存を待つタスク。\\n- **何が完了したか?** -- `completed`のタスク。完了時に後続タスクを自動的にアンブロックする。\\n\\n```\\n.tasks/\\n task_1.json {\\\"id\\\":1, \\\"status\\\":\\\"completed\\\"}\\n task_2.json {\\\"id\\\":2, \\\"blockedBy\\\":[1], \\\"status\\\":\\\"pending\\\"}\\n task_3.json {\\\"id\\\":3, \\\"blockedBy\\\":[1], \\\"status\\\":\\\"pending\\\"}\\n task_4.json {\\\"id\\\":4, \\\"blockedBy\\\":[2,3], \\\"status\\\":\\\"pending\\\"}\\n\\nタスクグラフ (DAG):\\n +----------+\\n +--> | task 2 | --+\\n | | pending | |\\n+----------+ +----------+ +--> +----------+\\n| task 1 | | task 4 |\\n| completed| --> +----------+ +--> | blocked |\\n+----------+ | task 3 | --+ +----------+\\n | pending |\\n +----------+\\n\\n順序: task 1 は 2 と 3 より先に完了する必要がある\\n並行: task 2 と 3 は同時に実行できる\\n依存: task 4 は 2 と 3 の両方を待つ\\nステータス: pending -> in_progress -> completed\\n```\\n\\nこのタスクグラフは s07 以降の全メカニズムの協調バックボーンとなる: バックグラウンド実行(s08)、マルチエージェントチーム(s09+)、worktree分離(s12)はすべてこの同じ構造を読み書きする。\\n\\n## 仕組み\\n\\n1. **TaskManager**: タスクごとに1つのJSONファイル、依存グラフ付きCRUD。\\n\\n```python\\nclass TaskManager:\\n def __init__(self, tasks_dir: Path):\\n self.dir = tasks_dir\\n self.dir.mkdir(exist_ok=True)\\n self._next_id = self._max_id() + 1\\n\\n def create(self, subject, description=\\\"\\\"):\\n task = {\\\"id\\\": self._next_id, \\\"subject\\\": subject,\\n \\\"status\\\": \\\"pending\\\", \\\"blockedBy\\\": [],\\n \\\"blocks\\\": [], \\\"owner\\\": \\\"\\\"}\\n self._save(task)\\n self._next_id += 1\\n return json.dumps(task, indent=2)\\n```\\n\\n2. **依存解除**: タスク完了時に、他タスクの`blockedBy`リストから完了IDを除去し、後続タスクをアンブロックする。\\n\\n```python\\ndef _clear_dependency(self, completed_id):\\n for f in self.dir.glob(\\\"task_*.json\\\"):\\n task = json.loads(f.read_text())\\n if completed_id in task.get(\\\"blockedBy\\\", []):\\n task[\\\"blockedBy\\\"].remove(completed_id)\\n self._save(task)\\n```\\n\\n3. **ステータス遷移 + 依存配線**: `update`がステータス変更と依存エッジを担う。\\n\\n```python\\ndef update(self, task_id, status=None,\\n add_blocked_by=None, add_blocks=None):\\n task = self._load(task_id)\\n if status:\\n task[\\\"status\\\"] = status\\n if status == \\\"completed\\\":\\n self._clear_dependency(task_id)\\n self._save(task)\\n```\\n\\n4. 4つのタスクツールをディスパッチマップに追加する。\\n\\n```python\\nTOOL_HANDLERS = {\\n # ...base tools...\\n \\\"task_create\\\": lambda **kw: TASKS.create(kw[\\\"subject\\\"]),\\n \\\"task_update\\\": lambda **kw: TASKS.update(kw[\\\"task_id\\\"], kw.get(\\\"status\\\")),\\n \\\"task_list\\\": lambda **kw: TASKS.list_all(),\\n \\\"task_get\\\": lambda **kw: TASKS.get(kw[\\\"task_id\\\"]),\\n}\\n```\\n\\ns07以降、タスクグラフがマルチステップ作業のデフォルト。s03のTodoは軽量な単一セッション用チェックリストとして残る。\\n\\n## s06からの変更点\\n\\n| コンポーネント | Before (s06) | After (s07) |\\n|---|---|---|\\n| Tools | 5 | 8 (`task_create/update/list/get`) |\\n| 計画モデル | フラットチェックリスト (メモリ) | 依存関係付きタスクグラフ (ディスク) |\\n| 関係 | なし | `blockedBy` + `blocks` エッジ |\\n| ステータス追跡 | 完了か未完了 | `pending` -> `in_progress` -> `completed` |\\n| 永続性 | 圧縮で消失 | 圧縮・再起動後も存続 |\\n\\n## 試してみる\\n\\n```sh\\ncd learn-claude-code\\npython agents/s07_task_system.py\\n```\\n\\n1. `Create 3 tasks: \\\"Setup project\\\", \\\"Write code\\\", \\\"Write tests\\\". Make them depend on each other in order.`\\n2. `List all tasks and show the dependency graph`\\n3. `Complete task 1 and then list tasks to see task 2 unblocked`\\n4. `Create a task board for refactoring: parse -> transform -> emit -> test, where transform and emit can run in parallel after parse`\\n\"\n },\n {\n \"version\": \"s08\",\n \"locale\": \"ja\",\n \"title\": \"s08: Background Tasks\",\n \"content\": \"# s08: Background Tasks\\n\\n`s01 > s02 > s03 > s04 > s05 > s06 | s07 > [ s08 ] s09 > s10 > s11 > s12`\\n\\n> *\\\"遅い操作はバックグラウンドへ、エージェントは次を考え続ける\\\"* -- デーモンスレッドがコマンド実行、完了後に通知を注入。\\n\\n## 問題\\n\\n一部のコマンドは数分かかる: `npm install`、`pytest`、`docker build`。ブロッキングループでは、モデルはサブプロセスの完了を待って座っている。ユーザーが「依存関係をインストールして、その間にconfigファイルを作って」と言っても、エージェントは並列ではなく逐次的に処理する。\\n\\n## 解決策\\n\\n```\\nMain thread Background thread\\n+-----------------+ +-----------------+\\n| agent loop | | subprocess runs |\\n| ... | | ... |\\n| [LLM call] <---+------- | enqueue(result) |\\n| ^drain queue | +-----------------+\\n+-----------------+\\n\\nTimeline:\\nAgent --[spawn A]--[spawn B]--[other work]----\\n | |\\n v v\\n [A runs] [B runs] (parallel)\\n | |\\n +-- results injected before next LLM call --+\\n```\\n\\n## 仕組み\\n\\n1. BackgroundManagerがスレッドセーフな通知キューでタスクを追跡する。\\n\\n```python\\nclass BackgroundManager:\\n def __init__(self):\\n self.tasks = {}\\n self._notification_queue = []\\n self._lock = threading.Lock()\\n```\\n\\n2. `run()`がデーモンスレッドを開始し、即座にリターンする。\\n\\n```python\\ndef run(self, command: str) -> str:\\n task_id = str(uuid.uuid4())[:8]\\n self.tasks[task_id] = {\\\"status\\\": \\\"running\\\", \\\"command\\\": command}\\n thread = threading.Thread(\\n target=self._execute, args=(task_id, command), daemon=True)\\n thread.start()\\n return f\\\"Background task {task_id} started\\\"\\n```\\n\\n3. サブプロセス完了時に、結果を通知キューへ。\\n\\n```python\\ndef _execute(self, task_id, command):\\n try:\\n r = subprocess.run(command, shell=True, cwd=WORKDIR,\\n capture_output=True, text=True, timeout=300)\\n output = (r.stdout + r.stderr).strip()[:50000]\\n except subprocess.TimeoutExpired:\\n output = \\\"Error: Timeout (300s)\\\"\\n with self._lock:\\n self._notification_queue.append({\\n \\\"task_id\\\": task_id, \\\"result\\\": output[:500]})\\n```\\n\\n4. エージェントループが各LLM呼び出しの前に通知をドレインする。\\n\\n```python\\ndef agent_loop(messages: list):\\n while True:\\n notifs = BG.drain_notifications()\\n if notifs:\\n notif_text = \\\"\\\\n\\\".join(\\n f\\\"[bg:{n['task_id']}] {n['result']}\\\" for n in notifs)\\n messages.append({\\\"role\\\": \\\"user\\\",\\n \\\"content\\\": f\\\"\\\\n{notif_text}\\\\n\\\"\\n f\\\" \\\"})\\n messages.append({\\\"role\\\": \\\"assistant\\\",\\n \\\"content\\\": \\\"Noted background results.\\\"})\\n response = client.messages.create(...)\\n```\\n\\nループはシングルスレッドのまま。サブプロセスI/Oだけが並列化される。\\n\\n## s07からの変更点\\n\\n| Component | Before (s07) | After (s08) |\\n|----------------|------------------|----------------------------|\\n| Tools | 8 | 6 (base + background_run + check)|\\n| Execution | Blocking only | Blocking + background threads|\\n| Notification | None | Queue drained per loop |\\n| Concurrency | None | Daemon threads |\\n\\n## 試してみる\\n\\n```sh\\ncd learn-claude-code\\npython agents/s08_background_tasks.py\\n```\\n\\n1. `Run \\\"sleep 5 && echo done\\\" in the background, then create a file while it runs`\\n2. `Start 3 background tasks: \\\"sleep 2\\\", \\\"sleep 4\\\", \\\"sleep 6\\\". Check their status.`\\n3. `Run pytest in the background and keep working on other things`\\n\"\n },\n {\n \"version\": \"s09\",\n \"locale\": \"ja\",\n \"title\": \"s09: Agent Teams\",\n \"content\": \"# s09: Agent Teams\\n\\n`s01 > s02 > s03 > s04 > s05 > s06 | s07 > s08 > [ s09 ] s10 > s11 > s12`\\n\\n> *\\\"一人で終わらないなら、チームメイトに任せる\\\"* -- 永続チームメイト + 非同期メールボックス。\\n\\n## 問題\\n\\nサブエージェント(s04)は使い捨てだ: 生成し、作業し、要約を返し、消滅する。アイデンティティもなく、呼び出し間の記憶もない。バックグラウンドタスク(s08)はシェルコマンドを実行するが、LLM誘導の意思決定はできない。\\n\\n本物のチームワークには: (1)単一プロンプトを超えて存続する永続エージェント、(2)アイデンティティとライフサイクル管理、(3)エージェント間の通信チャネルが必要だ。\\n\\n## 解決策\\n\\n```\\nTeammate lifecycle:\\n spawn -> WORKING -> IDLE -> WORKING -> ... -> SHUTDOWN\\n\\nCommunication:\\n .team/\\n config.json <- team roster + statuses\\n inbox/\\n alice.jsonl <- append-only, drain-on-read\\n bob.jsonl\\n lead.jsonl\\n\\n +--------+ send(\\\"alice\\\",\\\"bob\\\",\\\"...\\\") +--------+\\n | alice | -----------------------------> | bob |\\n | loop | bob.jsonl << {json_line} | loop |\\n +--------+ +--------+\\n ^ |\\n | BUS.read_inbox(\\\"alice\\\") |\\n +---- alice.jsonl -> read + drain ---------+\\n```\\n\\n## 仕組み\\n\\n1. TeammateManagerがconfig.jsonでチーム名簿を管理する。\\n\\n```python\\nclass TeammateManager:\\n def __init__(self, team_dir: Path):\\n self.dir = team_dir\\n self.dir.mkdir(exist_ok=True)\\n self.config_path = self.dir / \\\"config.json\\\"\\n self.config = self._load_config()\\n self.threads = {}\\n```\\n\\n2. `spawn()`がチームメイトを作成し、そのエージェントループをスレッドで開始する。\\n\\n```python\\ndef spawn(self, name: str, role: str, prompt: str) -> str:\\n member = {\\\"name\\\": name, \\\"role\\\": role, \\\"status\\\": \\\"working\\\"}\\n self.config[\\\"members\\\"].append(member)\\n self._save_config()\\n thread = threading.Thread(\\n target=self._teammate_loop,\\n args=(name, role, prompt), daemon=True)\\n thread.start()\\n return f\\\"Spawned teammate '{name}' (role: {role})\\\"\\n```\\n\\n3. MessageBus: 追記専用のJSONLインボックス。`send()`がJSON行を追記し、`read_inbox()`がすべて読み取ってドレインする。\\n\\n```python\\nclass MessageBus:\\n def send(self, sender, to, content, msg_type=\\\"message\\\", extra=None):\\n msg = {\\\"type\\\": msg_type, \\\"from\\\": sender,\\n \\\"content\\\": content, \\\"timestamp\\\": time.time()}\\n if extra:\\n msg.update(extra)\\n with open(self.dir / f\\\"{to}.jsonl\\\", \\\"a\\\") as f:\\n f.write(json.dumps(msg) + \\\"\\\\n\\\")\\n\\n def read_inbox(self, name):\\n path = self.dir / f\\\"{name}.jsonl\\\"\\n if not path.exists(): return \\\"[]\\\"\\n msgs = [json.loads(l) for l in path.read_text().strip().splitlines() if l]\\n path.write_text(\\\"\\\") # drain\\n return json.dumps(msgs, indent=2)\\n```\\n\\n4. 各チームメイトは各LLM呼び出しの前にインボックスを確認し、受信メッセージをコンテキストに注入する。\\n\\n```python\\ndef _teammate_loop(self, name, role, prompt):\\n messages = [{\\\"role\\\": \\\"user\\\", \\\"content\\\": prompt}]\\n for _ in range(50):\\n inbox = BUS.read_inbox(name)\\n if inbox != \\\"[]\\\":\\n messages.append({\\\"role\\\": \\\"user\\\",\\n \\\"content\\\": f\\\"{inbox} \\\"})\\n messages.append({\\\"role\\\": \\\"assistant\\\",\\n \\\"content\\\": \\\"Noted inbox messages.\\\"})\\n response = client.messages.create(...)\\n if response.stop_reason != \\\"tool_use\\\":\\n break\\n # execute tools, append results...\\n self._find_member(name)[\\\"status\\\"] = \\\"idle\\\"\\n```\\n\\n## s08からの変更点\\n\\n| Component | Before (s08) | After (s09) |\\n|----------------|------------------|----------------------------|\\n| Tools | 6 | 9 (+spawn/send/read_inbox) |\\n| Agents | Single | Lead + N teammates |\\n| Persistence | None | config.json + JSONL inboxes|\\n| Threads | Background cmds | Full agent loops per thread|\\n| Lifecycle | Fire-and-forget | idle -> working -> idle |\\n| Communication | None | message + broadcast |\\n\\n## 試してみる\\n\\n```sh\\ncd learn-claude-code\\npython agents/s09_agent_teams.py\\n```\\n\\n1. `Spawn alice (coder) and bob (tester). Have alice send bob a message.`\\n2. `Broadcast \\\"status update: phase 1 complete\\\" to all teammates`\\n3. `Check the lead inbox for any messages`\\n4. `/team`と入力してステータス付きのチーム名簿を確認する\\n5. `/inbox`と入力してリーダーのインボックスを手動確認する\\n\"\n },\n {\n \"version\": \"s10\",\n \"locale\": \"ja\",\n \"title\": \"s10: Team Protocols\",\n \"content\": \"# s10: Team Protocols\\n\\n`s01 > s02 > s03 > s04 > s05 > s06 | s07 > s08 > s09 > [ s10 ] s11 > s12`\\n\\n> *\\\"チームメイト間には統一の通信ルールが必要\\\"* -- 1つの request-response パターンが全交渉を駆動。\\n\\n## 問題\\n\\ns09ではチームメイトが作業し通信するが、構造化された協調がない:\\n\\n**シャットダウン**: スレッドを強制終了するとファイルが中途半端に書かれ、config.jsonが不正な状態になる。ハンドシェイクが必要 -- リーダーが要求し、チームメイトが承認(完了して退出)か拒否(作業継続)する。\\n\\n**プラン承認**: リーダーが「認証モジュールをリファクタリングして」と言うと、チームメイトは即座に開始する。リスクの高い変更では、実行前にリーダーが計画をレビューすべきだ。\\n\\n両方とも同じ構造: 一方がユニークIDを持つリクエストを送り、他方がそのIDで応答する。\\n\\n## 解決策\\n\\n```\\nShutdown Protocol Plan Approval Protocol\\n================== ======================\\n\\nLead Teammate Teammate Lead\\n | | | |\\n |--shutdown_req-->| |--plan_req------>|\\n | {req_id:\\\"abc\\\"} | | {req_id:\\\"xyz\\\"} |\\n | | | |\\n |<--shutdown_resp-| |<--plan_resp-----|\\n | {req_id:\\\"abc\\\", | | {req_id:\\\"xyz\\\", |\\n | approve:true} | | approve:true} |\\n\\nShared FSM:\\n [pending] --approve--> [approved]\\n [pending] --reject---> [rejected]\\n\\nTrackers:\\n shutdown_requests = {req_id: {target, status}}\\n plan_requests = {req_id: {from, plan, status}}\\n```\\n\\n## 仕組み\\n\\n1. リーダーがrequest_idを生成し、インボックス経由でシャットダウンを開始する。\\n\\n```python\\nshutdown_requests = {}\\n\\ndef handle_shutdown_request(teammate: str) -> str:\\n req_id = str(uuid.uuid4())[:8]\\n shutdown_requests[req_id] = {\\\"target\\\": teammate, \\\"status\\\": \\\"pending\\\"}\\n BUS.send(\\\"lead\\\", teammate, \\\"Please shut down gracefully.\\\",\\n \\\"shutdown_request\\\", {\\\"request_id\\\": req_id})\\n return f\\\"Shutdown request {req_id} sent (status: pending)\\\"\\n```\\n\\n2. チームメイトがリクエストを受信し、承認または拒否で応答する。\\n\\n```python\\nif tool_name == \\\"shutdown_response\\\":\\n req_id = args[\\\"request_id\\\"]\\n approve = args[\\\"approve\\\"]\\n shutdown_requests[req_id][\\\"status\\\"] = \\\"approved\\\" if approve else \\\"rejected\\\"\\n BUS.send(sender, \\\"lead\\\", args.get(\\\"reason\\\", \\\"\\\"),\\n \\\"shutdown_response\\\",\\n {\\\"request_id\\\": req_id, \\\"approve\\\": approve})\\n```\\n\\n3. プラン承認も同一パターン。チームメイトがプランを提出(request_idを生成)、リーダーがレビュー(同じrequest_idを参照)。\\n\\n```python\\nplan_requests = {}\\n\\ndef handle_plan_review(request_id, approve, feedback=\\\"\\\"):\\n req = plan_requests[request_id]\\n req[\\\"status\\\"] = \\\"approved\\\" if approve else \\\"rejected\\\"\\n BUS.send(\\\"lead\\\", req[\\\"from\\\"], feedback,\\n \\\"plan_approval_response\\\",\\n {\\\"request_id\\\": request_id, \\\"approve\\\": approve})\\n```\\n\\n1つのFSM、2つの応用。同じ`pending -> approved | rejected`状態機械が、あらゆるリクエスト-レスポンスプロトコルに適用できる。\\n\\n## s09からの変更点\\n\\n| Component | Before (s09) | After (s10) |\\n|----------------|------------------|------------------------------|\\n| Tools | 9 | 12 (+shutdown_req/resp +plan)|\\n| Shutdown | Natural exit only| Request-response handshake |\\n| Plan gating | None | Submit/review with approval |\\n| Correlation | None | request_id per request |\\n| FSM | None | pending -> approved/rejected |\\n\\n## 試してみる\\n\\n```sh\\ncd learn-claude-code\\npython agents/s10_team_protocols.py\\n```\\n\\n1. `Spawn alice as a coder. Then request her shutdown.`\\n2. `List teammates to see alice's status after shutdown approval`\\n3. `Spawn bob with a risky refactoring task. Review and reject his plan.`\\n4. `Spawn charlie, have him submit a plan, then approve it.`\\n5. `/team`と入力してステータスを監視する\\n\"\n },\n {\n \"version\": \"s11\",\n \"locale\": \"ja\",\n \"title\": \"s11: Autonomous Agents\",\n \"content\": \"# s11: Autonomous Agents\\n\\n`s01 > s02 > s03 > s04 > s05 > s06 | s07 > s08 > s09 > s10 > [ s11 ] s12`\\n\\n> *\\\"チームメイトが自らボードを見て、仕事を取る\\\"* -- リーダーが逐一割り振る必要はない。\\n\\n## 問題\\n\\ns09-s10では、チームメイトは明示的に指示された時のみ作業する。リーダーは各チームメイトを特定のプロンプトでspawnしなければならない。タスクボードに未割り当てのタスクが10個あっても、リーダーが手動で各タスクを割り当てる。これはスケールしない。\\n\\n真の自律性とは、チームメイトが自分で作業を見つけること: タスクボードをスキャンし、未確保のタスクを確保し、作業し、完了したら次を探す。\\n\\nもう1つの問題: コンテキスト圧縮(s06)後にエージェントが自分の正体を忘れる可能性がある。アイデンティティ再注入がこれを解決する。\\n\\n## 解決策\\n\\n```\\nTeammate lifecycle with idle cycle:\\n\\n+-------+\\n| spawn |\\n+---+---+\\n |\\n v\\n+-------+ tool_use +-------+\\n| WORK | <------------- | LLM |\\n+---+---+ +-------+\\n |\\n | stop_reason != tool_use (or idle tool called)\\n v\\n+--------+\\n| IDLE | poll every 5s for up to 60s\\n+---+----+\\n |\\n +---> check inbox --> message? ----------> WORK\\n |\\n +---> scan .tasks/ --> unclaimed? -------> claim -> WORK\\n |\\n +---> 60s timeout ----------------------> SHUTDOWN\\n\\nIdentity re-injection after compression:\\n if len(messages) <= 3:\\n messages.insert(0, identity_block)\\n```\\n\\n## 仕組み\\n\\n1. チームメイトのループはWORKとIDLEの2フェーズ。LLMがツール呼び出しを止めた時(または`idle`ツールを呼んだ時)、IDLEフェーズに入る。\\n\\n```python\\ndef _loop(self, name, role, prompt):\\n while True:\\n # -- WORK PHASE --\\n messages = [{\\\"role\\\": \\\"user\\\", \\\"content\\\": prompt}]\\n for _ in range(50):\\n response = client.messages.create(...)\\n if response.stop_reason != \\\"tool_use\\\":\\n break\\n # execute tools...\\n if idle_requested:\\n break\\n\\n # -- IDLE PHASE --\\n self._set_status(name, \\\"idle\\\")\\n resume = self._idle_poll(name, messages)\\n if not resume:\\n self._set_status(name, \\\"shutdown\\\")\\n return\\n self._set_status(name, \\\"working\\\")\\n```\\n\\n2. IDLEフェーズがインボックスとタスクボードをポーリングする。\\n\\n```python\\ndef _idle_poll(self, name, messages):\\n for _ in range(IDLE_TIMEOUT // POLL_INTERVAL): # 60s / 5s = 12\\n time.sleep(POLL_INTERVAL)\\n inbox = BUS.read_inbox(name)\\n if inbox:\\n messages.append({\\\"role\\\": \\\"user\\\",\\n \\\"content\\\": f\\\"{inbox} \\\"})\\n return True\\n unclaimed = scan_unclaimed_tasks()\\n if unclaimed:\\n claim_task(unclaimed[0][\\\"id\\\"], name)\\n messages.append({\\\"role\\\": \\\"user\\\",\\n \\\"content\\\": f\\\"Task #{unclaimed[0]['id']}: \\\"\\n f\\\"{unclaimed[0]['subject']} \\\"})\\n return True\\n return False # timeout -> shutdown\\n```\\n\\n3. タスクボードスキャン: pendingかつ未割り当てかつブロックされていないタスクを探す。\\n\\n```python\\ndef scan_unclaimed_tasks() -> list:\\n unclaimed = []\\n for f in sorted(TASKS_DIR.glob(\\\"task_*.json\\\")):\\n task = json.loads(f.read_text())\\n if (task.get(\\\"status\\\") == \\\"pending\\\"\\n and not task.get(\\\"owner\\\")\\n and not task.get(\\\"blockedBy\\\")):\\n unclaimed.append(task)\\n return unclaimed\\n```\\n\\n4. アイデンティティ再注入: コンテキストが短すぎる(圧縮が起きた)場合にアイデンティティブロックを挿入する。\\n\\n```python\\nif len(messages) <= 3:\\n messages.insert(0, {\\\"role\\\": \\\"user\\\",\\n \\\"content\\\": f\\\"You are '{name}', role: {role}, \\\"\\n f\\\"team: {team_name}. Continue your work. \\\"})\\n messages.insert(1, {\\\"role\\\": \\\"assistant\\\",\\n \\\"content\\\": f\\\"I am {name}. Continuing.\\\"})\\n```\\n\\n## s10からの変更点\\n\\n| Component | Before (s10) | After (s11) |\\n|----------------|------------------|----------------------------|\\n| Tools | 12 | 14 (+idle, +claim_task) |\\n| Autonomy | Lead-directed | Self-organizing |\\n| Idle phase | None | Poll inbox + task board |\\n| Task claiming | Manual only | Auto-claim unclaimed tasks |\\n| Identity | System prompt | + re-injection after compress|\\n| Timeout | None | 60s idle -> auto shutdown |\\n\\n## 試してみる\\n\\n```sh\\ncd learn-claude-code\\npython agents/s11_autonomous_agents.py\\n```\\n\\n1. `Create 3 tasks on the board, then spawn alice and bob. Watch them auto-claim.`\\n2. `Spawn a coder teammate and let it find work from the task board itself`\\n3. `Create tasks with dependencies. Watch teammates respect the blocked order.`\\n4. `/tasks`と入力してオーナー付きのタスクボードを確認する\\n5. `/team`と入力して誰が作業中でアイドルかを監視する\\n\"\n },\n {\n \"version\": \"s12\",\n \"locale\": \"ja\",\n \"title\": \"s12: Worktree + Task Isolation\",\n \"content\": \"# s12: Worktree + Task Isolation\\n\\n`s01 > s02 > s03 > s04 > s05 > s06 | s07 > s08 > s09 > s10 > s11 > [ s12 ]`\\n\\n> *\\\"各自のディレクトリで作業し、互いに干渉しない\\\"* -- タスクは目標を管理、worktree はディレクトリを管理、IDで紐付け。\\n\\n## 問題\\n\\ns11までにエージェントはタスクを自律的に確保して完了できるようになった。しかし全タスクが1つの共有ディレクトリで走る。2つのエージェントが同時に異なるモジュールをリファクタリングすると衝突する: 片方が`config.py`を編集し、もう片方も`config.py`を編集し、未コミットの変更が混ざり合い、どちらもクリーンにロールバックできない。\\n\\nタスクボードは*何をやるか*を追跡するが、*どこでやるか*には関知しない。解決策: 各タスクに専用のgit worktreeディレクトリを与える。タスクが目標を管理し、worktreeが実行コンテキストを管理する。タスクIDで紐付ける。\\n\\n## 解決策\\n\\n```\\nControl plane (.tasks/) Execution plane (.worktrees/)\\n+------------------+ +------------------------+\\n| task_1.json | | auth-refactor/ |\\n| status: in_progress <------> branch: wt/auth-refactor\\n| worktree: \\\"auth-refactor\\\" | task_id: 1 |\\n+------------------+ +------------------------+\\n| task_2.json | | ui-login/ |\\n| status: pending <------> branch: wt/ui-login\\n| worktree: \\\"ui-login\\\" | task_id: 2 |\\n+------------------+ +------------------------+\\n |\\n index.json (worktree registry)\\n events.jsonl (lifecycle log)\\n\\nState machines:\\n Task: pending -> in_progress -> completed\\n Worktree: absent -> active -> removed | kept\\n```\\n\\n## 仕組み\\n\\n1. **タスクを作成する。** まず目標を永続化する。\\n\\n```python\\nTASKS.create(\\\"Implement auth refactor\\\")\\n# -> .tasks/task_1.json status=pending worktree=\\\"\\\"\\n```\\n\\n2. **worktreeを作成してタスクに紐付ける。** `task_id`を渡すと、タスクが自動的に`in_progress`に遷移する。\\n\\n```python\\nWORKTREES.create(\\\"auth-refactor\\\", task_id=1)\\n# -> git worktree add -b wt/auth-refactor .worktrees/auth-refactor HEAD\\n# -> index.json gets new entry, task_1.json gets worktree=\\\"auth-refactor\\\"\\n```\\n\\n紐付けは両側に状態を書き込む:\\n\\n```python\\ndef bind_worktree(self, task_id, worktree):\\n task = self._load(task_id)\\n task[\\\"worktree\\\"] = worktree\\n if task[\\\"status\\\"] == \\\"pending\\\":\\n task[\\\"status\\\"] = \\\"in_progress\\\"\\n self._save(task)\\n```\\n\\n3. **worktree内でコマンドを実行する。** `cwd`が分離ディレクトリを指す。\\n\\n```python\\nsubprocess.run(command, shell=True, cwd=worktree_path,\\n capture_output=True, text=True, timeout=300)\\n```\\n\\n4. **終了処理。** 2つの選択肢:\\n - `worktree_keep(name)` -- ディレクトリを保持する。\\n - `worktree_remove(name, complete_task=True)` -- ディレクトリを削除し、紐付けられたタスクを完了し、イベントを発行する。1回の呼び出しで後片付けと完了を処理する。\\n\\n```python\\ndef remove(self, name, force=False, complete_task=False):\\n self._run_git([\\\"worktree\\\", \\\"remove\\\", wt[\\\"path\\\"]])\\n if complete_task and wt.get(\\\"task_id\\\") is not None:\\n self.tasks.update(wt[\\\"task_id\\\"], status=\\\"completed\\\")\\n self.tasks.unbind_worktree(wt[\\\"task_id\\\"])\\n self.events.emit(\\\"task.completed\\\", ...)\\n```\\n\\n5. **イベントストリーム。** ライフサイクルの各ステップが`.worktrees/events.jsonl`に記録される:\\n\\n```json\\n{\\n \\\"event\\\": \\\"worktree.remove.after\\\",\\n \\\"task\\\": {\\\"id\\\": 1, \\\"status\\\": \\\"completed\\\"},\\n \\\"worktree\\\": {\\\"name\\\": \\\"auth-refactor\\\", \\\"status\\\": \\\"removed\\\"},\\n \\\"ts\\\": 1730000000\\n}\\n```\\n\\n発行されるイベント: `worktree.create.before/after/failed`, `worktree.remove.before/after/failed`, `worktree.keep`, `task.completed`。\\n\\nクラッシュ後も`.tasks/` + `.worktrees/index.json`から状態を再構築できる。会話メモリは揮発性だが、ファイル状態は永続的だ。\\n\\n## s11からの変更点\\n\\n| Component | Before (s11) | After (s12) |\\n|--------------------|----------------------------|----------------------------------------------|\\n| Coordination | Task board (owner/status) | Task board + explicit worktree binding |\\n| Execution scope | Shared directory | Task-scoped isolated directory |\\n| Recoverability | Task status only | Task status + worktree index |\\n| Teardown | Task completion | Task completion + explicit keep/remove |\\n| Lifecycle visibility | Implicit in logs | Explicit events in `.worktrees/events.jsonl` |\\n\\n## 試してみる\\n\\n```sh\\ncd learn-claude-code\\npython agents/s12_worktree_task_isolation.py\\n```\\n\\n1. `Create tasks for backend auth and frontend login page, then list tasks.`\\n2. `Create worktree \\\"auth-refactor\\\" for task 1, then bind task 2 to a new worktree \\\"ui-login\\\".`\\n3. `Run \\\"git status --short\\\" in worktree \\\"auth-refactor\\\".`\\n4. `Keep worktree \\\"ui-login\\\", then list worktrees and inspect events.`\\n5. `Remove worktree \\\"auth-refactor\\\" with complete_task=true, then list tasks/worktrees/events.`\\n\"\n }\n]"
},
{
"path": "web/src/data/generated/versions.json",
"content": "{\n \"versions\": [\n {\n \"id\": \"s01\",\n \"filename\": \"s01_agent_loop.py\",\n \"title\": \"The Agent Loop\",\n \"subtitle\": \"Bash is All You Need\",\n \"loc\": 84,\n \"tools\": [\n \"bash\"\n ],\n \"newTools\": [\n \"bash\"\n ],\n \"coreAddition\": \"Single-tool agent loop\",\n \"keyInsight\": \"The minimal agent kernel is a while loop + one tool\",\n \"classes\": [],\n \"functions\": [\n {\n \"name\": \"run_bash\",\n \"signature\": \"def run_bash(command: str)\",\n \"startLine\": 53\n },\n {\n \"name\": \"agent_loop\",\n \"signature\": \"def agent_loop(messages: list)\",\n \"startLine\": 67\n }\n ],\n \"layer\": \"tools\",\n \"source\": \"#!/usr/bin/env python3\\n\\\"\\\"\\\"\\ns01_agent_loop.py - The Agent Loop\\n\\nThe entire secret of an AI coding agent in one pattern:\\n\\n while stop_reason == \\\"tool_use\\\":\\n response = LLM(messages, tools)\\n execute tools\\n append results\\n\\n +----------+ +-------+ +---------+\\n | User | ---> | LLM | ---> | Tool |\\n | prompt | | | | execute |\\n +----------+ +---+---+ +----+----+\\n ^ |\\n | tool_result |\\n +---------------+\\n (loop continues)\\n\\nThis is the core loop: feed tool results back to the model\\nuntil the model decides to stop. Production agents layer\\npolicy, hooks, and lifecycle controls on top.\\n\\\"\\\"\\\"\\n\\nimport os\\nimport subprocess\\n\\nfrom anthropic import Anthropic\\nfrom dotenv import load_dotenv\\n\\nload_dotenv(override=True)\\n\\nif os.getenv(\\\"ANTHROPIC_BASE_URL\\\"):\\n os.environ.pop(\\\"ANTHROPIC_AUTH_TOKEN\\\", None)\\n\\nclient = Anthropic(base_url=os.getenv(\\\"ANTHROPIC_BASE_URL\\\"))\\nMODEL = os.environ[\\\"MODEL_ID\\\"]\\n\\nSYSTEM = f\\\"You are a coding agent at {os.getcwd()}. Use bash to solve tasks. Act, don't explain.\\\"\\n\\nTOOLS = [{\\n \\\"name\\\": \\\"bash\\\",\\n \\\"description\\\": \\\"Run a shell command.\\\",\\n \\\"input_schema\\\": {\\n \\\"type\\\": \\\"object\\\",\\n \\\"properties\\\": {\\\"command\\\": {\\\"type\\\": \\\"string\\\"}},\\n \\\"required\\\": [\\\"command\\\"],\\n },\\n}]\\n\\n\\ndef run_bash(command: str) -> str:\\n dangerous = [\\\"rm -rf /\\\", \\\"sudo\\\", \\\"shutdown\\\", \\\"reboot\\\", \\\"> /dev/\\\"]\\n if any(d in command for d in dangerous):\\n return \\\"Error: Dangerous command blocked\\\"\\n try:\\n r = subprocess.run(command, shell=True, cwd=os.getcwd(),\\n capture_output=True, text=True, timeout=120)\\n out = (r.stdout + r.stderr).strip()\\n return out[:50000] if out else \\\"(no output)\\\"\\n except subprocess.TimeoutExpired:\\n return \\\"Error: Timeout (120s)\\\"\\n\\n\\n# -- The core pattern: a while loop that calls tools until the model stops --\\ndef agent_loop(messages: list):\\n while True:\\n response = client.messages.create(\\n model=MODEL, system=SYSTEM, messages=messages,\\n tools=TOOLS, max_tokens=8000,\\n )\\n # Append assistant turn\\n messages.append({\\\"role\\\": \\\"assistant\\\", \\\"content\\\": response.content})\\n # If the model didn't call a tool, we're done\\n if response.stop_reason != \\\"tool_use\\\":\\n return\\n # Execute each tool call, collect results\\n results = []\\n for block in response.content:\\n if block.type == \\\"tool_use\\\":\\n print(f\\\"\\\\033[33m$ {block.input['command']}\\\\033[0m\\\")\\n output = run_bash(block.input[\\\"command\\\"])\\n print(output[:200])\\n results.append({\\\"type\\\": \\\"tool_result\\\", \\\"tool_use_id\\\": block.id,\\n \\\"content\\\": output})\\n messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": results})\\n\\n\\nif __name__ == \\\"__main__\\\":\\n history = []\\n while True:\\n try:\\n query = input(\\\"\\\\033[36ms01 >> \\\\033[0m\\\")\\n except (EOFError, KeyboardInterrupt):\\n break\\n if query.strip().lower() in (\\\"q\\\", \\\"exit\\\", \\\"\\\"):\\n break\\n history.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": query})\\n agent_loop(history)\\n response_content = history[-1][\\\"content\\\"]\\n if isinstance(response_content, list):\\n for block in response_content:\\n if hasattr(block, \\\"text\\\"):\\n print(block.text)\\n print()\\n\"\n },\n {\n \"id\": \"s02\",\n \"filename\": \"s02_tool_use.py\",\n \"title\": \"Tools\",\n \"subtitle\": \"One Handler Per Tool\",\n \"loc\": 115,\n \"tools\": [\n \"bash\",\n \"read_file\",\n \"write_file\",\n \"edit_file\"\n ],\n \"newTools\": [\n \"read_file\",\n \"write_file\",\n \"edit_file\"\n ],\n \"coreAddition\": \"Tool dispatch map\",\n \"keyInsight\": \"The loop stays the same; new tools register into the dispatch map\",\n \"classes\": [],\n \"functions\": [\n {\n \"name\": \"safe_path\",\n \"signature\": \"def safe_path(p: str)\",\n \"startLine\": 40\n },\n {\n \"name\": \"run_bash\",\n \"signature\": \"def run_bash(command: str)\",\n \"startLine\": 47\n },\n {\n \"name\": \"run_read\",\n \"signature\": \"def run_read(path: str, limit: int = None)\",\n \"startLine\": 60\n },\n {\n \"name\": \"run_write\",\n \"signature\": \"def run_write(path: str, content: str)\",\n \"startLine\": 71\n },\n {\n \"name\": \"run_edit\",\n \"signature\": \"def run_edit(path: str, old_text: str, new_text: str)\",\n \"startLine\": 81\n },\n {\n \"name\": \"agent_loop\",\n \"signature\": \"def agent_loop(messages: list)\",\n \"startLine\": 113\n }\n ],\n \"layer\": \"tools\",\n \"source\": \"#!/usr/bin/env python3\\n\\\"\\\"\\\"\\ns02_tool_use.py - Tools\\n\\nThe agent loop from s01 didn't change. We just added tools to the array\\nand a dispatch map to route calls.\\n\\n +----------+ +-------+ +------------------+\\n | User | ---> | LLM | ---> | Tool Dispatch |\\n | prompt | | | | { |\\n +----------+ +---+---+ | bash: run_bash |\\n ^ | read: run_read |\\n | | write: run_wr |\\n +----------+ edit: run_edit |\\n tool_result| } |\\n +------------------+\\n\\nKey insight: \\\"The loop didn't change at all. I just added tools.\\\"\\n\\\"\\\"\\\"\\n\\nimport os\\nimport subprocess\\nfrom pathlib import Path\\n\\nfrom anthropic import Anthropic\\nfrom dotenv import load_dotenv\\n\\nload_dotenv(override=True)\\n\\nif os.getenv(\\\"ANTHROPIC_BASE_URL\\\"):\\n os.environ.pop(\\\"ANTHROPIC_AUTH_TOKEN\\\", None)\\n\\nWORKDIR = Path.cwd()\\nclient = Anthropic(base_url=os.getenv(\\\"ANTHROPIC_BASE_URL\\\"))\\nMODEL = os.environ[\\\"MODEL_ID\\\"]\\n\\nSYSTEM = f\\\"You are a coding agent at {WORKDIR}. Use tools to solve tasks. Act, don't explain.\\\"\\n\\n\\ndef safe_path(p: str) -> Path:\\n path = (WORKDIR / p).resolve()\\n if not path.is_relative_to(WORKDIR):\\n raise ValueError(f\\\"Path escapes workspace: {p}\\\")\\n return path\\n\\n\\ndef run_bash(command: str) -> str:\\n dangerous = [\\\"rm -rf /\\\", \\\"sudo\\\", \\\"shutdown\\\", \\\"reboot\\\", \\\"> /dev/\\\"]\\n if any(d in command for d in dangerous):\\n return \\\"Error: Dangerous command blocked\\\"\\n try:\\n r = subprocess.run(command, shell=True, cwd=WORKDIR,\\n capture_output=True, text=True, timeout=120)\\n out = (r.stdout + r.stderr).strip()\\n return out[:50000] if out else \\\"(no output)\\\"\\n except subprocess.TimeoutExpired:\\n return \\\"Error: Timeout (120s)\\\"\\n\\n\\ndef run_read(path: str, limit: int = None) -> str:\\n try:\\n text = safe_path(path).read_text()\\n lines = text.splitlines()\\n if limit and limit < len(lines):\\n lines = lines[:limit] + [f\\\"... ({len(lines) - limit} more lines)\\\"]\\n return \\\"\\\\n\\\".join(lines)[:50000]\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\n\\ndef run_write(path: str, content: str) -> str:\\n try:\\n fp = safe_path(path)\\n fp.parent.mkdir(parents=True, exist_ok=True)\\n fp.write_text(content)\\n return f\\\"Wrote {len(content)} bytes to {path}\\\"\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\n\\ndef run_edit(path: str, old_text: str, new_text: str) -> str:\\n try:\\n fp = safe_path(path)\\n content = fp.read_text()\\n if old_text not in content:\\n return f\\\"Error: Text not found in {path}\\\"\\n fp.write_text(content.replace(old_text, new_text, 1))\\n return f\\\"Edited {path}\\\"\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\n\\n# -- The dispatch map: {tool_name: handler} --\\nTOOL_HANDLERS = {\\n \\\"bash\\\": lambda **kw: run_bash(kw[\\\"command\\\"]),\\n \\\"read_file\\\": lambda **kw: run_read(kw[\\\"path\\\"], kw.get(\\\"limit\\\")),\\n \\\"write_file\\\": lambda **kw: run_write(kw[\\\"path\\\"], kw[\\\"content\\\"]),\\n \\\"edit_file\\\": lambda **kw: run_edit(kw[\\\"path\\\"], kw[\\\"old_text\\\"], kw[\\\"new_text\\\"]),\\n}\\n\\nTOOLS = [\\n {\\\"name\\\": \\\"bash\\\", \\\"description\\\": \\\"Run a shell command.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"command\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"command\\\"]}},\\n {\\\"name\\\": \\\"read_file\\\", \\\"description\\\": \\\"Read file contents.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"limit\\\": {\\\"type\\\": \\\"integer\\\"}}, \\\"required\\\": [\\\"path\\\"]}},\\n {\\\"name\\\": \\\"write_file\\\", \\\"description\\\": \\\"Write content to file.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"content\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\", \\\"content\\\"]}},\\n {\\\"name\\\": \\\"edit_file\\\", \\\"description\\\": \\\"Replace exact text in file.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"old_text\\\": {\\\"type\\\": \\\"string\\\"}, \\\"new_text\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\", \\\"old_text\\\", \\\"new_text\\\"]}},\\n]\\n\\n\\ndef agent_loop(messages: list):\\n while True:\\n response = client.messages.create(\\n model=MODEL, system=SYSTEM, messages=messages,\\n tools=TOOLS, max_tokens=8000,\\n )\\n messages.append({\\\"role\\\": \\\"assistant\\\", \\\"content\\\": response.content})\\n if response.stop_reason != \\\"tool_use\\\":\\n return\\n results = []\\n for block in response.content:\\n if block.type == \\\"tool_use\\\":\\n handler = TOOL_HANDLERS.get(block.name)\\n output = handler(**block.input) if handler else f\\\"Unknown tool: {block.name}\\\"\\n print(f\\\"> {block.name}: {output[:200]}\\\")\\n results.append({\\\"type\\\": \\\"tool_result\\\", \\\"tool_use_id\\\": block.id, \\\"content\\\": output})\\n messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": results})\\n\\n\\nif __name__ == \\\"__main__\\\":\\n history = []\\n while True:\\n try:\\n query = input(\\\"\\\\033[36ms02 >> \\\\033[0m\\\")\\n except (EOFError, KeyboardInterrupt):\\n break\\n if query.strip().lower() in (\\\"q\\\", \\\"exit\\\", \\\"\\\"):\\n break\\n history.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": query})\\n agent_loop(history)\\n print()\\n\"\n },\n {\n \"id\": \"s03\",\n \"filename\": \"s03_todo_write.py\",\n \"title\": \"TodoWrite\",\n \"subtitle\": \"Plan Before You Act\",\n \"loc\": 171,\n \"tools\": [\n \"bash\",\n \"read_file\",\n \"write_file\",\n \"edit_file\",\n \"todo\"\n ],\n \"newTools\": [\n \"todo\"\n ],\n \"coreAddition\": \"TodoManager + nag reminder\",\n \"keyInsight\": \"An agent without a plan drifts; list the steps first, then execute\",\n \"classes\": [\n {\n \"name\": \"TodoManager\",\n \"startLine\": 51,\n \"endLine\": 87\n }\n ],\n \"functions\": [\n {\n \"name\": \"safe_path\",\n \"signature\": \"def safe_path(p: str)\",\n \"startLine\": 92\n },\n {\n \"name\": \"run_bash\",\n \"signature\": \"def run_bash(command: str)\",\n \"startLine\": 98\n },\n {\n \"name\": \"run_read\",\n \"signature\": \"def run_read(path: str, limit: int = None)\",\n \"startLine\": 110\n },\n {\n \"name\": \"run_write\",\n \"signature\": \"def run_write(path: str, content: str)\",\n \"startLine\": 119\n },\n {\n \"name\": \"run_edit\",\n \"signature\": \"def run_edit(path: str, old_text: str, new_text: str)\",\n \"startLine\": 128\n },\n {\n \"name\": \"agent_loop\",\n \"signature\": \"def agent_loop(messages: list)\",\n \"startLine\": 163\n }\n ],\n \"layer\": \"planning\",\n \"source\": \"#!/usr/bin/env python3\\n\\\"\\\"\\\"\\ns03_todo_write.py - TodoWrite\\n\\nThe model tracks its own progress via a TodoManager. A nag reminder\\nforces it to keep updating when it forgets.\\n\\n +----------+ +-------+ +---------+\\n | User | ---> | LLM | ---> | Tools |\\n | prompt | | | | + todo |\\n +----------+ +---+---+ +----+----+\\n ^ |\\n | tool_result |\\n +---------------+\\n |\\n +-----------+-----------+\\n | TodoManager state |\\n | [ ] task A |\\n | [>] task B <- doing |\\n | [x] task C |\\n +-----------------------+\\n |\\n if rounds_since_todo >= 3:\\n inject \\n\\nKey insight: \\\"The agent can track its own progress -- and I can see it.\\\"\\n\\\"\\\"\\\"\\n\\nimport os\\nimport subprocess\\nfrom pathlib import Path\\n\\nfrom anthropic import Anthropic\\nfrom dotenv import load_dotenv\\n\\nload_dotenv(override=True)\\n\\nif os.getenv(\\\"ANTHROPIC_BASE_URL\\\"):\\n os.environ.pop(\\\"ANTHROPIC_AUTH_TOKEN\\\", None)\\n\\nWORKDIR = Path.cwd()\\nclient = Anthropic(base_url=os.getenv(\\\"ANTHROPIC_BASE_URL\\\"))\\nMODEL = os.environ[\\\"MODEL_ID\\\"]\\n\\nSYSTEM = f\\\"\\\"\\\"You are a coding agent at {WORKDIR}.\\nUse the todo tool to plan multi-step tasks. Mark in_progress before starting, completed when done.\\nPrefer tools over prose.\\\"\\\"\\\"\\n\\n\\n# -- TodoManager: structured state the LLM writes to --\\nclass TodoManager:\\n def __init__(self):\\n self.items = []\\n\\n def update(self, items: list) -> str:\\n if len(items) > 20:\\n raise ValueError(\\\"Max 20 todos allowed\\\")\\n validated = []\\n in_progress_count = 0\\n for i, item in enumerate(items):\\n text = str(item.get(\\\"text\\\", \\\"\\\")).strip()\\n status = str(item.get(\\\"status\\\", \\\"pending\\\")).lower()\\n item_id = str(item.get(\\\"id\\\", str(i + 1)))\\n if not text:\\n raise ValueError(f\\\"Item {item_id}: text required\\\")\\n if status not in (\\\"pending\\\", \\\"in_progress\\\", \\\"completed\\\"):\\n raise ValueError(f\\\"Item {item_id}: invalid status '{status}'\\\")\\n if status == \\\"in_progress\\\":\\n in_progress_count += 1\\n validated.append({\\\"id\\\": item_id, \\\"text\\\": text, \\\"status\\\": status})\\n if in_progress_count > 1:\\n raise ValueError(\\\"Only one task can be in_progress at a time\\\")\\n self.items = validated\\n return self.render()\\n\\n def render(self) -> str:\\n if not self.items:\\n return \\\"No todos.\\\"\\n lines = []\\n for item in self.items:\\n marker = {\\\"pending\\\": \\\"[ ]\\\", \\\"in_progress\\\": \\\"[>]\\\", \\\"completed\\\": \\\"[x]\\\"}[item[\\\"status\\\"]]\\n lines.append(f\\\"{marker} #{item['id']}: {item['text']}\\\")\\n done = sum(1 for t in self.items if t[\\\"status\\\"] == \\\"completed\\\")\\n lines.append(f\\\"\\\\n({done}/{len(self.items)} completed)\\\")\\n return \\\"\\\\n\\\".join(lines)\\n\\n\\nTODO = TodoManager()\\n\\n\\n# -- Tool implementations --\\ndef safe_path(p: str) -> Path:\\n path = (WORKDIR / p).resolve()\\n if not path.is_relative_to(WORKDIR):\\n raise ValueError(f\\\"Path escapes workspace: {p}\\\")\\n return path\\n\\ndef run_bash(command: str) -> str:\\n dangerous = [\\\"rm -rf /\\\", \\\"sudo\\\", \\\"shutdown\\\", \\\"reboot\\\", \\\"> /dev/\\\"]\\n if any(d in command for d in dangerous):\\n return \\\"Error: Dangerous command blocked\\\"\\n try:\\n r = subprocess.run(command, shell=True, cwd=WORKDIR,\\n capture_output=True, text=True, timeout=120)\\n out = (r.stdout + r.stderr).strip()\\n return out[:50000] if out else \\\"(no output)\\\"\\n except subprocess.TimeoutExpired:\\n return \\\"Error: Timeout (120s)\\\"\\n\\ndef run_read(path: str, limit: int = None) -> str:\\n try:\\n lines = safe_path(path).read_text().splitlines()\\n if limit and limit < len(lines):\\n lines = lines[:limit] + [f\\\"... ({len(lines) - limit} more)\\\"]\\n return \\\"\\\\n\\\".join(lines)[:50000]\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\ndef run_write(path: str, content: str) -> str:\\n try:\\n fp = safe_path(path)\\n fp.parent.mkdir(parents=True, exist_ok=True)\\n fp.write_text(content)\\n return f\\\"Wrote {len(content)} bytes\\\"\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\ndef run_edit(path: str, old_text: str, new_text: str) -> str:\\n try:\\n fp = safe_path(path)\\n content = fp.read_text()\\n if old_text not in content:\\n return f\\\"Error: Text not found in {path}\\\"\\n fp.write_text(content.replace(old_text, new_text, 1))\\n return f\\\"Edited {path}\\\"\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\n\\nTOOL_HANDLERS = {\\n \\\"bash\\\": lambda **kw: run_bash(kw[\\\"command\\\"]),\\n \\\"read_file\\\": lambda **kw: run_read(kw[\\\"path\\\"], kw.get(\\\"limit\\\")),\\n \\\"write_file\\\": lambda **kw: run_write(kw[\\\"path\\\"], kw[\\\"content\\\"]),\\n \\\"edit_file\\\": lambda **kw: run_edit(kw[\\\"path\\\"], kw[\\\"old_text\\\"], kw[\\\"new_text\\\"]),\\n \\\"todo\\\": lambda **kw: TODO.update(kw[\\\"items\\\"]),\\n}\\n\\nTOOLS = [\\n {\\\"name\\\": \\\"bash\\\", \\\"description\\\": \\\"Run a shell command.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"command\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"command\\\"]}},\\n {\\\"name\\\": \\\"read_file\\\", \\\"description\\\": \\\"Read file contents.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"limit\\\": {\\\"type\\\": \\\"integer\\\"}}, \\\"required\\\": [\\\"path\\\"]}},\\n {\\\"name\\\": \\\"write_file\\\", \\\"description\\\": \\\"Write content to file.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"content\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\", \\\"content\\\"]}},\\n {\\\"name\\\": \\\"edit_file\\\", \\\"description\\\": \\\"Replace exact text in file.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"old_text\\\": {\\\"type\\\": \\\"string\\\"}, \\\"new_text\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\", \\\"old_text\\\", \\\"new_text\\\"]}},\\n {\\\"name\\\": \\\"todo\\\", \\\"description\\\": \\\"Update task list. Track progress on multi-step tasks.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"items\\\": {\\\"type\\\": \\\"array\\\", \\\"items\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"id\\\": {\\\"type\\\": \\\"string\\\"}, \\\"text\\\": {\\\"type\\\": \\\"string\\\"}, \\\"status\\\": {\\\"type\\\": \\\"string\\\", \\\"enum\\\": [\\\"pending\\\", \\\"in_progress\\\", \\\"completed\\\"]}}, \\\"required\\\": [\\\"id\\\", \\\"text\\\", \\\"status\\\"]}}}, \\\"required\\\": [\\\"items\\\"]}},\\n]\\n\\n\\n# -- Agent loop with nag reminder injection --\\ndef agent_loop(messages: list):\\n rounds_since_todo = 0\\n while True:\\n # Nag reminder is injected below, alongside tool results\\n response = client.messages.create(\\n model=MODEL, system=SYSTEM, messages=messages,\\n tools=TOOLS, max_tokens=8000,\\n )\\n messages.append({\\\"role\\\": \\\"assistant\\\", \\\"content\\\": response.content})\\n if response.stop_reason != \\\"tool_use\\\":\\n return\\n results = []\\n used_todo = False\\n for block in response.content:\\n if block.type == \\\"tool_use\\\":\\n handler = TOOL_HANDLERS.get(block.name)\\n try:\\n output = handler(**block.input) if handler else f\\\"Unknown tool: {block.name}\\\"\\n except Exception as e:\\n output = f\\\"Error: {e}\\\"\\n print(f\\\"> {block.name}: {str(output)[:200]}\\\")\\n results.append({\\\"type\\\": \\\"tool_result\\\", \\\"tool_use_id\\\": block.id, \\\"content\\\": str(output)})\\n if block.name == \\\"todo\\\":\\n used_todo = True\\n rounds_since_todo = 0 if used_todo else rounds_since_todo + 1\\n if rounds_since_todo >= 3:\\n results.insert(0, {\\\"type\\\": \\\"text\\\", \\\"text\\\": \\\"Update your todos. \\\"})\\n messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": results})\\n\\n\\nif __name__ == \\\"__main__\\\":\\n history = []\\n while True:\\n try:\\n query = input(\\\"\\\\033[36ms03 >> \\\\033[0m\\\")\\n except (EOFError, KeyboardInterrupt):\\n break\\n if query.strip().lower() in (\\\"q\\\", \\\"exit\\\", \\\"\\\"):\\n break\\n history.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": query})\\n agent_loop(history)\\n print()\\n\"\n },\n {\n \"id\": \"s04\",\n \"filename\": \"s04_subagent.py\",\n \"title\": \"Subagents\",\n \"subtitle\": \"Clean Context Per Subtask\",\n \"loc\": 146,\n \"tools\": [\n \"bash\",\n \"read_file\",\n \"write_file\",\n \"edit_file\",\n \"task\"\n ],\n \"newTools\": [\n \"task\"\n ],\n \"coreAddition\": \"Subagent spawn with isolated messages[]\",\n \"keyInsight\": \"Subagents use independent messages[], keeping the main conversation clean\",\n \"classes\": [],\n \"functions\": [\n {\n \"name\": \"safe_path\",\n \"signature\": \"def safe_path(p: str)\",\n \"startLine\": 46\n },\n {\n \"name\": \"run_bash\",\n \"signature\": \"def run_bash(command: str)\",\n \"startLine\": 52\n },\n {\n \"name\": \"run_read\",\n \"signature\": \"def run_read(path: str, limit: int = None)\",\n \"startLine\": 64\n },\n {\n \"name\": \"run_write\",\n \"signature\": \"def run_write(path: str, content: str)\",\n \"startLine\": 73\n },\n {\n \"name\": \"run_edit\",\n \"signature\": \"def run_edit(path: str, old_text: str, new_text: str)\",\n \"startLine\": 82\n },\n {\n \"name\": \"run_subagent\",\n \"signature\": \"def run_subagent(prompt: str)\",\n \"startLine\": 115\n },\n {\n \"name\": \"agent_loop\",\n \"signature\": \"def agent_loop(messages: list)\",\n \"startLine\": 143\n }\n ],\n \"layer\": \"planning\",\n \"source\": \"#!/usr/bin/env python3\\n\\\"\\\"\\\"\\ns04_subagent.py - Subagents\\n\\nSpawn a child agent with fresh messages=[]. The child works in its own\\ncontext, sharing the filesystem, then returns only a summary to the parent.\\n\\n Parent agent Subagent\\n +------------------+ +------------------+\\n | messages=[...] | | messages=[] | <-- fresh\\n | | dispatch | |\\n | tool: task | ---------->| while tool_use: |\\n | prompt=\\\"...\\\" | | call tools |\\n | description=\\\"\\\" | | append results |\\n | | summary | |\\n | result = \\\"...\\\" | <--------- | return last text |\\n +------------------+ +------------------+\\n |\\n Parent context stays clean.\\n Subagent context is discarded.\\n\\nKey insight: \\\"Process isolation gives context isolation for free.\\\"\\n\\\"\\\"\\\"\\n\\nimport os\\nimport subprocess\\nfrom pathlib import Path\\n\\nfrom anthropic import Anthropic\\nfrom dotenv import load_dotenv\\n\\nload_dotenv(override=True)\\n\\nif os.getenv(\\\"ANTHROPIC_BASE_URL\\\"):\\n os.environ.pop(\\\"ANTHROPIC_AUTH_TOKEN\\\", None)\\n\\nWORKDIR = Path.cwd()\\nclient = Anthropic(base_url=os.getenv(\\\"ANTHROPIC_BASE_URL\\\"))\\nMODEL = os.environ[\\\"MODEL_ID\\\"]\\n\\nSYSTEM = f\\\"You are a coding agent at {WORKDIR}. Use the task tool to delegate exploration or subtasks.\\\"\\nSUBAGENT_SYSTEM = f\\\"You are a coding subagent at {WORKDIR}. Complete the given task, then summarize your findings.\\\"\\n\\n\\n# -- Tool implementations shared by parent and child --\\ndef safe_path(p: str) -> Path:\\n path = (WORKDIR / p).resolve()\\n if not path.is_relative_to(WORKDIR):\\n raise ValueError(f\\\"Path escapes workspace: {p}\\\")\\n return path\\n\\ndef run_bash(command: str) -> str:\\n dangerous = [\\\"rm -rf /\\\", \\\"sudo\\\", \\\"shutdown\\\", \\\"reboot\\\", \\\"> /dev/\\\"]\\n if any(d in command for d in dangerous):\\n return \\\"Error: Dangerous command blocked\\\"\\n try:\\n r = subprocess.run(command, shell=True, cwd=WORKDIR,\\n capture_output=True, text=True, timeout=120)\\n out = (r.stdout + r.stderr).strip()\\n return out[:50000] if out else \\\"(no output)\\\"\\n except subprocess.TimeoutExpired:\\n return \\\"Error: Timeout (120s)\\\"\\n\\ndef run_read(path: str, limit: int = None) -> str:\\n try:\\n lines = safe_path(path).read_text().splitlines()\\n if limit and limit < len(lines):\\n lines = lines[:limit] + [f\\\"... ({len(lines) - limit} more)\\\"]\\n return \\\"\\\\n\\\".join(lines)[:50000]\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\ndef run_write(path: str, content: str) -> str:\\n try:\\n fp = safe_path(path)\\n fp.parent.mkdir(parents=True, exist_ok=True)\\n fp.write_text(content)\\n return f\\\"Wrote {len(content)} bytes\\\"\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\ndef run_edit(path: str, old_text: str, new_text: str) -> str:\\n try:\\n fp = safe_path(path)\\n content = fp.read_text()\\n if old_text not in content:\\n return f\\\"Error: Text not found in {path}\\\"\\n fp.write_text(content.replace(old_text, new_text, 1))\\n return f\\\"Edited {path}\\\"\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\n\\nTOOL_HANDLERS = {\\n \\\"bash\\\": lambda **kw: run_bash(kw[\\\"command\\\"]),\\n \\\"read_file\\\": lambda **kw: run_read(kw[\\\"path\\\"], kw.get(\\\"limit\\\")),\\n \\\"write_file\\\": lambda **kw: run_write(kw[\\\"path\\\"], kw[\\\"content\\\"]),\\n \\\"edit_file\\\": lambda **kw: run_edit(kw[\\\"path\\\"], kw[\\\"old_text\\\"], kw[\\\"new_text\\\"]),\\n}\\n\\n# Child gets all base tools except task (no recursive spawning)\\nCHILD_TOOLS = [\\n {\\\"name\\\": \\\"bash\\\", \\\"description\\\": \\\"Run a shell command.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"command\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"command\\\"]}},\\n {\\\"name\\\": \\\"read_file\\\", \\\"description\\\": \\\"Read file contents.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"limit\\\": {\\\"type\\\": \\\"integer\\\"}}, \\\"required\\\": [\\\"path\\\"]}},\\n {\\\"name\\\": \\\"write_file\\\", \\\"description\\\": \\\"Write content to file.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"content\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\", \\\"content\\\"]}},\\n {\\\"name\\\": \\\"edit_file\\\", \\\"description\\\": \\\"Replace exact text in file.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"old_text\\\": {\\\"type\\\": \\\"string\\\"}, \\\"new_text\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\", \\\"old_text\\\", \\\"new_text\\\"]}},\\n]\\n\\n\\n# -- Subagent: fresh context, filtered tools, summary-only return --\\ndef run_subagent(prompt: str) -> str:\\n sub_messages = [{\\\"role\\\": \\\"user\\\", \\\"content\\\": prompt}] # fresh context\\n for _ in range(30): # safety limit\\n response = client.messages.create(\\n model=MODEL, system=SUBAGENT_SYSTEM, messages=sub_messages,\\n tools=CHILD_TOOLS, max_tokens=8000,\\n )\\n sub_messages.append({\\\"role\\\": \\\"assistant\\\", \\\"content\\\": response.content})\\n if response.stop_reason != \\\"tool_use\\\":\\n break\\n results = []\\n for block in response.content:\\n if block.type == \\\"tool_use\\\":\\n handler = TOOL_HANDLERS.get(block.name)\\n output = handler(**block.input) if handler else f\\\"Unknown tool: {block.name}\\\"\\n results.append({\\\"type\\\": \\\"tool_result\\\", \\\"tool_use_id\\\": block.id, \\\"content\\\": str(output)[:50000]})\\n sub_messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": results})\\n # Only the final text returns to the parent -- child context is discarded\\n return \\\"\\\".join(b.text for b in response.content if hasattr(b, \\\"text\\\")) or \\\"(no summary)\\\"\\n\\n\\n# -- Parent tools: base tools + task dispatcher --\\nPARENT_TOOLS = CHILD_TOOLS + [\\n {\\\"name\\\": \\\"task\\\", \\\"description\\\": \\\"Spawn a subagent with fresh context. It shares the filesystem but not conversation history.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"prompt\\\": {\\\"type\\\": \\\"string\\\"}, \\\"description\\\": {\\\"type\\\": \\\"string\\\", \\\"description\\\": \\\"Short description of the task\\\"}}, \\\"required\\\": [\\\"prompt\\\"]}},\\n]\\n\\n\\ndef agent_loop(messages: list):\\n while True:\\n response = client.messages.create(\\n model=MODEL, system=SYSTEM, messages=messages,\\n tools=PARENT_TOOLS, max_tokens=8000,\\n )\\n messages.append({\\\"role\\\": \\\"assistant\\\", \\\"content\\\": response.content})\\n if response.stop_reason != \\\"tool_use\\\":\\n return\\n results = []\\n for block in response.content:\\n if block.type == \\\"tool_use\\\":\\n if block.name == \\\"task\\\":\\n desc = block.input.get(\\\"description\\\", \\\"subtask\\\")\\n print(f\\\"> task ({desc}): {block.input['prompt'][:80]}\\\")\\n output = run_subagent(block.input[\\\"prompt\\\"])\\n else:\\n handler = TOOL_HANDLERS.get(block.name)\\n output = handler(**block.input) if handler else f\\\"Unknown tool: {block.name}\\\"\\n print(f\\\" {str(output)[:200]}\\\")\\n results.append({\\\"type\\\": \\\"tool_result\\\", \\\"tool_use_id\\\": block.id, \\\"content\\\": str(output)})\\n messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": results})\\n\\n\\nif __name__ == \\\"__main__\\\":\\n history = []\\n while True:\\n try:\\n query = input(\\\"\\\\033[36ms04 >> \\\\033[0m\\\")\\n except (EOFError, KeyboardInterrupt):\\n break\\n if query.strip().lower() in (\\\"q\\\", \\\"exit\\\", \\\"\\\"):\\n break\\n history.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": query})\\n agent_loop(history)\\n print()\\n\"\n },\n {\n \"id\": \"s05\",\n \"filename\": \"s05_skill_loading.py\",\n \"title\": \"Skills\",\n \"subtitle\": \"Load on Demand\",\n \"loc\": 182,\n \"tools\": [\n \"bash\",\n \"read_file\",\n \"write_file\",\n \"edit_file\",\n \"load_skill\"\n ],\n \"newTools\": [\n \"load_skill\"\n ],\n \"coreAddition\": \"SkillLoader + two-layer injection\",\n \"keyInsight\": \"Inject knowledge via tool_result when needed, not upfront in the system prompt\",\n \"classes\": [\n {\n \"name\": \"SkillLoader\",\n \"startLine\": 57,\n \"endLine\": 105\n }\n ],\n \"functions\": [\n {\n \"name\": \"safe_path\",\n \"signature\": \"def safe_path(p: str)\",\n \"startLine\": 117\n },\n {\n \"name\": \"run_bash\",\n \"signature\": \"def run_bash(command: str)\",\n \"startLine\": 123\n },\n {\n \"name\": \"run_read\",\n \"signature\": \"def run_read(path: str, limit: int = None)\",\n \"startLine\": 135\n },\n {\n \"name\": \"run_write\",\n \"signature\": \"def run_write(path: str, content: str)\",\n \"startLine\": 144\n },\n {\n \"name\": \"run_edit\",\n \"signature\": \"def run_edit(path: str, old_text: str, new_text: str)\",\n \"startLine\": 153\n },\n {\n \"name\": \"agent_loop\",\n \"signature\": \"def agent_loop(messages: list)\",\n \"startLine\": 187\n }\n ],\n \"layer\": \"planning\",\n \"source\": \"#!/usr/bin/env python3\\n\\\"\\\"\\\"\\ns05_skill_loading.py - Skills\\n\\nTwo-layer skill injection that avoids bloating the system prompt:\\n\\n Layer 1 (cheap): skill names in system prompt (~100 tokens/skill)\\n Layer 2 (on demand): full skill body in tool_result\\n\\n skills/\\n pdf/\\n SKILL.md <-- frontmatter (name, description) + body\\n code-review/\\n SKILL.md\\n\\n System prompt:\\n +--------------------------------------+\\n | You are a coding agent. |\\n | Skills available: |\\n | - pdf: Process PDF files... | <-- Layer 1: metadata only\\n | - code-review: Review code... |\\n +--------------------------------------+\\n\\n When model calls load_skill(\\\"pdf\\\"):\\n +--------------------------------------+\\n | tool_result: |\\n | |\\n | Full PDF processing instructions | <-- Layer 2: full body\\n | Step 1: ... |\\n | Step 2: ... |\\n | |\\n +--------------------------------------+\\n\\nKey insight: \\\"Don't put everything in the system prompt. Load on demand.\\\"\\n\\\"\\\"\\\"\\n\\nimport os\\nimport re\\nimport subprocess\\nfrom pathlib import Path\\n\\nfrom anthropic import Anthropic\\nfrom dotenv import load_dotenv\\n\\nload_dotenv(override=True)\\n\\nif os.getenv(\\\"ANTHROPIC_BASE_URL\\\"):\\n os.environ.pop(\\\"ANTHROPIC_AUTH_TOKEN\\\", None)\\n\\nWORKDIR = Path.cwd()\\nclient = Anthropic(base_url=os.getenv(\\\"ANTHROPIC_BASE_URL\\\"))\\nMODEL = os.environ[\\\"MODEL_ID\\\"]\\nSKILLS_DIR = WORKDIR / \\\"skills\\\"\\n\\n\\n# -- SkillLoader: scan skills//SKILL.md with YAML frontmatter --\\nclass SkillLoader:\\n def __init__(self, skills_dir: Path):\\n self.skills_dir = skills_dir\\n self.skills = {}\\n self._load_all()\\n\\n def _load_all(self):\\n if not self.skills_dir.exists():\\n return\\n for f in sorted(self.skills_dir.rglob(\\\"SKILL.md\\\")):\\n text = f.read_text()\\n meta, body = self._parse_frontmatter(text)\\n name = meta.get(\\\"name\\\", f.parent.name)\\n self.skills[name] = {\\\"meta\\\": meta, \\\"body\\\": body, \\\"path\\\": str(f)}\\n\\n def _parse_frontmatter(self, text: str) -> tuple:\\n \\\"\\\"\\\"Parse YAML frontmatter between --- delimiters.\\\"\\\"\\\"\\n match = re.match(r\\\"^---\\\\n(.*?)\\\\n---\\\\n(.*)\\\", text, re.DOTALL)\\n if not match:\\n return {}, text\\n meta = {}\\n for line in match.group(1).strip().splitlines():\\n if \\\":\\\" in line:\\n key, val = line.split(\\\":\\\", 1)\\n meta[key.strip()] = val.strip()\\n return meta, match.group(2).strip()\\n\\n def get_descriptions(self) -> str:\\n \\\"\\\"\\\"Layer 1: short descriptions for the system prompt.\\\"\\\"\\\"\\n if not self.skills:\\n return \\\"(no skills available)\\\"\\n lines = []\\n for name, skill in self.skills.items():\\n desc = skill[\\\"meta\\\"].get(\\\"description\\\", \\\"No description\\\")\\n tags = skill[\\\"meta\\\"].get(\\\"tags\\\", \\\"\\\")\\n line = f\\\" - {name}: {desc}\\\"\\n if tags:\\n line += f\\\" [{tags}]\\\"\\n lines.append(line)\\n return \\\"\\\\n\\\".join(lines)\\n\\n def get_content(self, name: str) -> str:\\n \\\"\\\"\\\"Layer 2: full skill body returned in tool_result.\\\"\\\"\\\"\\n skill = self.skills.get(name)\\n if not skill:\\n return f\\\"Error: Unknown skill '{name}'. Available: {', '.join(self.skills.keys())}\\\"\\n return f\\\"\\\\n{skill['body']}\\\\n \\\"\\n\\n\\nSKILL_LOADER = SkillLoader(SKILLS_DIR)\\n\\n# Layer 1: skill metadata injected into system prompt\\nSYSTEM = f\\\"\\\"\\\"You are a coding agent at {WORKDIR}.\\nUse load_skill to access specialized knowledge before tackling unfamiliar topics.\\n\\nSkills available:\\n{SKILL_LOADER.get_descriptions()}\\\"\\\"\\\"\\n\\n\\n# -- Tool implementations --\\ndef safe_path(p: str) -> Path:\\n path = (WORKDIR / p).resolve()\\n if not path.is_relative_to(WORKDIR):\\n raise ValueError(f\\\"Path escapes workspace: {p}\\\")\\n return path\\n\\ndef run_bash(command: str) -> str:\\n dangerous = [\\\"rm -rf /\\\", \\\"sudo\\\", \\\"shutdown\\\", \\\"reboot\\\", \\\"> /dev/\\\"]\\n if any(d in command for d in dangerous):\\n return \\\"Error: Dangerous command blocked\\\"\\n try:\\n r = subprocess.run(command, shell=True, cwd=WORKDIR,\\n capture_output=True, text=True, timeout=120)\\n out = (r.stdout + r.stderr).strip()\\n return out[:50000] if out else \\\"(no output)\\\"\\n except subprocess.TimeoutExpired:\\n return \\\"Error: Timeout (120s)\\\"\\n\\ndef run_read(path: str, limit: int = None) -> str:\\n try:\\n lines = safe_path(path).read_text().splitlines()\\n if limit and limit < len(lines):\\n lines = lines[:limit] + [f\\\"... ({len(lines) - limit} more)\\\"]\\n return \\\"\\\\n\\\".join(lines)[:50000]\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\ndef run_write(path: str, content: str) -> str:\\n try:\\n fp = safe_path(path)\\n fp.parent.mkdir(parents=True, exist_ok=True)\\n fp.write_text(content)\\n return f\\\"Wrote {len(content)} bytes\\\"\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\ndef run_edit(path: str, old_text: str, new_text: str) -> str:\\n try:\\n fp = safe_path(path)\\n content = fp.read_text()\\n if old_text not in content:\\n return f\\\"Error: Text not found in {path}\\\"\\n fp.write_text(content.replace(old_text, new_text, 1))\\n return f\\\"Edited {path}\\\"\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\n\\nTOOL_HANDLERS = {\\n \\\"bash\\\": lambda **kw: run_bash(kw[\\\"command\\\"]),\\n \\\"read_file\\\": lambda **kw: run_read(kw[\\\"path\\\"], kw.get(\\\"limit\\\")),\\n \\\"write_file\\\": lambda **kw: run_write(kw[\\\"path\\\"], kw[\\\"content\\\"]),\\n \\\"edit_file\\\": lambda **kw: run_edit(kw[\\\"path\\\"], kw[\\\"old_text\\\"], kw[\\\"new_text\\\"]),\\n \\\"load_skill\\\": lambda **kw: SKILL_LOADER.get_content(kw[\\\"name\\\"]),\\n}\\n\\nTOOLS = [\\n {\\\"name\\\": \\\"bash\\\", \\\"description\\\": \\\"Run a shell command.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"command\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"command\\\"]}},\\n {\\\"name\\\": \\\"read_file\\\", \\\"description\\\": \\\"Read file contents.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"limit\\\": {\\\"type\\\": \\\"integer\\\"}}, \\\"required\\\": [\\\"path\\\"]}},\\n {\\\"name\\\": \\\"write_file\\\", \\\"description\\\": \\\"Write content to file.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"content\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\", \\\"content\\\"]}},\\n {\\\"name\\\": \\\"edit_file\\\", \\\"description\\\": \\\"Replace exact text in file.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"old_text\\\": {\\\"type\\\": \\\"string\\\"}, \\\"new_text\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\", \\\"old_text\\\", \\\"new_text\\\"]}},\\n {\\\"name\\\": \\\"load_skill\\\", \\\"description\\\": \\\"Load specialized knowledge by name.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"name\\\": {\\\"type\\\": \\\"string\\\", \\\"description\\\": \\\"Skill name to load\\\"}}, \\\"required\\\": [\\\"name\\\"]}},\\n]\\n\\n\\ndef agent_loop(messages: list):\\n while True:\\n response = client.messages.create(\\n model=MODEL, system=SYSTEM, messages=messages,\\n tools=TOOLS, max_tokens=8000,\\n )\\n messages.append({\\\"role\\\": \\\"assistant\\\", \\\"content\\\": response.content})\\n if response.stop_reason != \\\"tool_use\\\":\\n return\\n results = []\\n for block in response.content:\\n if block.type == \\\"tool_use\\\":\\n handler = TOOL_HANDLERS.get(block.name)\\n try:\\n output = handler(**block.input) if handler else f\\\"Unknown tool: {block.name}\\\"\\n except Exception as e:\\n output = f\\\"Error: {e}\\\"\\n print(f\\\"> {block.name}: {str(output)[:200]}\\\")\\n results.append({\\\"type\\\": \\\"tool_result\\\", \\\"tool_use_id\\\": block.id, \\\"content\\\": str(output)})\\n messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": results})\\n\\n\\nif __name__ == \\\"__main__\\\":\\n history = []\\n while True:\\n try:\\n query = input(\\\"\\\\033[36ms05 >> \\\\033[0m\\\")\\n except (EOFError, KeyboardInterrupt):\\n break\\n if query.strip().lower() in (\\\"q\\\", \\\"exit\\\", \\\"\\\"):\\n break\\n history.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": query})\\n agent_loop(history)\\n print()\\n\"\n },\n {\n \"id\": \"s06\",\n \"filename\": \"s06_context_compact.py\",\n \"title\": \"Compact\",\n \"subtitle\": \"Three-Layer Compression\",\n \"loc\": 200,\n \"tools\": [\n \"bash\",\n \"read_file\",\n \"write_file\",\n \"edit_file\",\n \"compact\"\n ],\n \"newTools\": [\n \"compact\"\n ],\n \"coreAddition\": \"micro-compact + auto-compact + archival\",\n \"keyInsight\": \"Context will fill up; three-layer compression strategy enables infinite sessions\",\n \"classes\": [],\n \"functions\": [\n {\n \"name\": \"estimate_tokens\",\n \"signature\": \"def estimate_tokens(messages: list)\",\n \"startLine\": 61\n },\n {\n \"name\": \"micro_compact\",\n \"signature\": \"def micro_compact(messages: list)\",\n \"startLine\": 67\n },\n {\n \"name\": \"auto_compact\",\n \"signature\": \"def auto_compact(messages: list)\",\n \"startLine\": 97\n },\n {\n \"name\": \"safe_path\",\n \"signature\": \"def safe_path(p: str)\",\n \"startLine\": 124\n },\n {\n \"name\": \"run_bash\",\n \"signature\": \"def run_bash(command: str)\",\n \"startLine\": 130\n },\n {\n \"name\": \"run_read\",\n \"signature\": \"def run_read(path: str, limit: int = None)\",\n \"startLine\": 142\n },\n {\n \"name\": \"run_write\",\n \"signature\": \"def run_write(path: str, content: str)\",\n \"startLine\": 151\n },\n {\n \"name\": \"run_edit\",\n \"signature\": \"def run_edit(path: str, old_text: str, new_text: str)\",\n \"startLine\": 160\n },\n {\n \"name\": \"agent_loop\",\n \"signature\": \"def agent_loop(messages: list)\",\n \"startLine\": 194\n }\n ],\n \"layer\": \"memory\",\n \"source\": \"#!/usr/bin/env python3\\n\\\"\\\"\\\"\\ns06_context_compact.py - Compact\\n\\nThree-layer compression pipeline so the agent can work forever:\\n\\n Every turn:\\n +------------------+\\n | Tool call result |\\n +------------------+\\n |\\n v\\n [Layer 1: micro_compact] (silent, every turn)\\n Replace tool_result content older than last 3\\n with \\\"[Previous: used {tool_name}]\\\"\\n |\\n v\\n [Check: tokens > 50000?]\\n | |\\n no yes\\n | |\\n v v\\n continue [Layer 2: auto_compact]\\n Save full transcript to .transcripts/\\n Ask LLM to summarize conversation.\\n Replace all messages with [summary].\\n |\\n v\\n [Layer 3: compact tool]\\n Model calls compact -> immediate summarization.\\n Same as auto, triggered manually.\\n\\nKey insight: \\\"The agent can forget strategically and keep working forever.\\\"\\n\\\"\\\"\\\"\\n\\nimport json\\nimport os\\nimport subprocess\\nimport time\\nfrom pathlib import Path\\n\\nfrom anthropic import Anthropic\\nfrom dotenv import load_dotenv\\n\\nload_dotenv(override=True)\\n\\nif os.getenv(\\\"ANTHROPIC_BASE_URL\\\"):\\n os.environ.pop(\\\"ANTHROPIC_AUTH_TOKEN\\\", None)\\n\\nWORKDIR = Path.cwd()\\nclient = Anthropic(base_url=os.getenv(\\\"ANTHROPIC_BASE_URL\\\"))\\nMODEL = os.environ[\\\"MODEL_ID\\\"]\\n\\nSYSTEM = f\\\"You are a coding agent at {WORKDIR}. Use tools to solve tasks.\\\"\\n\\nTHRESHOLD = 50000\\nTRANSCRIPT_DIR = WORKDIR / \\\".transcripts\\\"\\nKEEP_RECENT = 3\\n\\n\\ndef estimate_tokens(messages: list) -> int:\\n \\\"\\\"\\\"Rough token count: ~4 chars per token.\\\"\\\"\\\"\\n return len(str(messages)) // 4\\n\\n\\n# -- Layer 1: micro_compact - replace old tool results with placeholders --\\ndef micro_compact(messages: list) -> list:\\n # Collect (msg_index, part_index, tool_result_dict) for all tool_result entries\\n tool_results = []\\n for msg_idx, msg in enumerate(messages):\\n if msg[\\\"role\\\"] == \\\"user\\\" and isinstance(msg.get(\\\"content\\\"), list):\\n for part_idx, part in enumerate(msg[\\\"content\\\"]):\\n if isinstance(part, dict) and part.get(\\\"type\\\") == \\\"tool_result\\\":\\n tool_results.append((msg_idx, part_idx, part))\\n if len(tool_results) <= KEEP_RECENT:\\n return messages\\n # Find tool_name for each result by matching tool_use_id in prior assistant messages\\n tool_name_map = {}\\n for msg in messages:\\n if msg[\\\"role\\\"] == \\\"assistant\\\":\\n content = msg.get(\\\"content\\\", [])\\n if isinstance(content, list):\\n for block in content:\\n if hasattr(block, \\\"type\\\") and block.type == \\\"tool_use\\\":\\n tool_name_map[block.id] = block.name\\n # Clear old results (keep last KEEP_RECENT)\\n to_clear = tool_results[:-KEEP_RECENT]\\n for _, _, result in to_clear:\\n if isinstance(result.get(\\\"content\\\"), str) and len(result[\\\"content\\\"]) > 100:\\n tool_id = result.get(\\\"tool_use_id\\\", \\\"\\\")\\n tool_name = tool_name_map.get(tool_id, \\\"unknown\\\")\\n result[\\\"content\\\"] = f\\\"[Previous: used {tool_name}]\\\"\\n return messages\\n\\n\\n# -- Layer 2: auto_compact - save transcript, summarize, replace messages --\\ndef auto_compact(messages: list) -> list:\\n # Save full transcript to disk\\n TRANSCRIPT_DIR.mkdir(exist_ok=True)\\n transcript_path = TRANSCRIPT_DIR / f\\\"transcript_{int(time.time())}.jsonl\\\"\\n with open(transcript_path, \\\"w\\\") as f:\\n for msg in messages:\\n f.write(json.dumps(msg, default=str) + \\\"\\\\n\\\")\\n print(f\\\"[transcript saved: {transcript_path}]\\\")\\n # Ask LLM to summarize\\n conversation_text = json.dumps(messages, default=str)[:80000]\\n response = client.messages.create(\\n model=MODEL,\\n messages=[{\\\"role\\\": \\\"user\\\", \\\"content\\\":\\n \\\"Summarize this conversation for continuity. Include: \\\"\\n \\\"1) What was accomplished, 2) Current state, 3) Key decisions made. \\\"\\n \\\"Be concise but preserve critical details.\\\\n\\\\n\\\" + conversation_text}],\\n max_tokens=2000,\\n )\\n summary = response.content[0].text\\n # Replace all messages with compressed summary\\n return [\\n {\\\"role\\\": \\\"user\\\", \\\"content\\\": f\\\"[Conversation compressed. Transcript: {transcript_path}]\\\\n\\\\n{summary}\\\"},\\n {\\\"role\\\": \\\"assistant\\\", \\\"content\\\": \\\"Understood. I have the context from the summary. Continuing.\\\"},\\n ]\\n\\n\\n# -- Tool implementations --\\ndef safe_path(p: str) -> Path:\\n path = (WORKDIR / p).resolve()\\n if not path.is_relative_to(WORKDIR):\\n raise ValueError(f\\\"Path escapes workspace: {p}\\\")\\n return path\\n\\ndef run_bash(command: str) -> str:\\n dangerous = [\\\"rm -rf /\\\", \\\"sudo\\\", \\\"shutdown\\\", \\\"reboot\\\", \\\"> /dev/\\\"]\\n if any(d in command for d in dangerous):\\n return \\\"Error: Dangerous command blocked\\\"\\n try:\\n r = subprocess.run(command, shell=True, cwd=WORKDIR,\\n capture_output=True, text=True, timeout=120)\\n out = (r.stdout + r.stderr).strip()\\n return out[:50000] if out else \\\"(no output)\\\"\\n except subprocess.TimeoutExpired:\\n return \\\"Error: Timeout (120s)\\\"\\n\\ndef run_read(path: str, limit: int = None) -> str:\\n try:\\n lines = safe_path(path).read_text().splitlines()\\n if limit and limit < len(lines):\\n lines = lines[:limit] + [f\\\"... ({len(lines) - limit} more)\\\"]\\n return \\\"\\\\n\\\".join(lines)[:50000]\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\ndef run_write(path: str, content: str) -> str:\\n try:\\n fp = safe_path(path)\\n fp.parent.mkdir(parents=True, exist_ok=True)\\n fp.write_text(content)\\n return f\\\"Wrote {len(content)} bytes\\\"\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\ndef run_edit(path: str, old_text: str, new_text: str) -> str:\\n try:\\n fp = safe_path(path)\\n content = fp.read_text()\\n if old_text not in content:\\n return f\\\"Error: Text not found in {path}\\\"\\n fp.write_text(content.replace(old_text, new_text, 1))\\n return f\\\"Edited {path}\\\"\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\n\\nTOOL_HANDLERS = {\\n \\\"bash\\\": lambda **kw: run_bash(kw[\\\"command\\\"]),\\n \\\"read_file\\\": lambda **kw: run_read(kw[\\\"path\\\"], kw.get(\\\"limit\\\")),\\n \\\"write_file\\\": lambda **kw: run_write(kw[\\\"path\\\"], kw[\\\"content\\\"]),\\n \\\"edit_file\\\": lambda **kw: run_edit(kw[\\\"path\\\"], kw[\\\"old_text\\\"], kw[\\\"new_text\\\"]),\\n \\\"compact\\\": lambda **kw: \\\"Manual compression requested.\\\",\\n}\\n\\nTOOLS = [\\n {\\\"name\\\": \\\"bash\\\", \\\"description\\\": \\\"Run a shell command.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"command\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"command\\\"]}},\\n {\\\"name\\\": \\\"read_file\\\", \\\"description\\\": \\\"Read file contents.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"limit\\\": {\\\"type\\\": \\\"integer\\\"}}, \\\"required\\\": [\\\"path\\\"]}},\\n {\\\"name\\\": \\\"write_file\\\", \\\"description\\\": \\\"Write content to file.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"content\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\", \\\"content\\\"]}},\\n {\\\"name\\\": \\\"edit_file\\\", \\\"description\\\": \\\"Replace exact text in file.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"old_text\\\": {\\\"type\\\": \\\"string\\\"}, \\\"new_text\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\", \\\"old_text\\\", \\\"new_text\\\"]}},\\n {\\\"name\\\": \\\"compact\\\", \\\"description\\\": \\\"Trigger manual conversation compression.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"focus\\\": {\\\"type\\\": \\\"string\\\", \\\"description\\\": \\\"What to preserve in the summary\\\"}}}},\\n]\\n\\n\\ndef agent_loop(messages: list):\\n while True:\\n # Layer 1: micro_compact before each LLM call\\n micro_compact(messages)\\n # Layer 2: auto_compact if token estimate exceeds threshold\\n if estimate_tokens(messages) > THRESHOLD:\\n print(\\\"[auto_compact triggered]\\\")\\n messages[:] = auto_compact(messages)\\n response = client.messages.create(\\n model=MODEL, system=SYSTEM, messages=messages,\\n tools=TOOLS, max_tokens=8000,\\n )\\n messages.append({\\\"role\\\": \\\"assistant\\\", \\\"content\\\": response.content})\\n if response.stop_reason != \\\"tool_use\\\":\\n return\\n results = []\\n manual_compact = False\\n for block in response.content:\\n if block.type == \\\"tool_use\\\":\\n if block.name == \\\"compact\\\":\\n manual_compact = True\\n output = \\\"Compressing...\\\"\\n else:\\n handler = TOOL_HANDLERS.get(block.name)\\n try:\\n output = handler(**block.input) if handler else f\\\"Unknown tool: {block.name}\\\"\\n except Exception as e:\\n output = f\\\"Error: {e}\\\"\\n print(f\\\"> {block.name}: {str(output)[:200]}\\\")\\n results.append({\\\"type\\\": \\\"tool_result\\\", \\\"tool_use_id\\\": block.id, \\\"content\\\": str(output)})\\n messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": results})\\n # Layer 3: manual compact triggered by the compact tool\\n if manual_compact:\\n print(\\\"[manual compact]\\\")\\n messages[:] = auto_compact(messages)\\n\\n\\nif __name__ == \\\"__main__\\\":\\n history = []\\n while True:\\n try:\\n query = input(\\\"\\\\033[36ms06 >> \\\\033[0m\\\")\\n except (EOFError, KeyboardInterrupt):\\n break\\n if query.strip().lower() in (\\\"q\\\", \\\"exit\\\", \\\"\\\"):\\n break\\n history.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": query})\\n agent_loop(history)\\n print()\\n\"\n },\n {\n \"id\": \"s07\",\n \"filename\": \"s07_task_system.py\",\n \"title\": \"Tasks\",\n \"subtitle\": \"Task Graph + Dependencies\",\n \"loc\": 202,\n \"tools\": [\n \"bash\",\n \"read_file\",\n \"write_file\",\n \"edit_file\",\n \"task_create\",\n \"task_update\",\n \"task_list\",\n \"task_get\"\n ],\n \"newTools\": [\n \"task_create\",\n \"task_update\",\n \"task_list\",\n \"task_get\"\n ],\n \"coreAddition\": \"TaskManager with file-based state + dependency graph\",\n \"keyInsight\": \"A file-based task graph with ordering, parallelism, and dependencies -- the coordination backbone for multi-agent work\",\n \"classes\": [\n {\n \"name\": \"TaskManager\",\n \"startLine\": 46,\n \"endLine\": 125\n }\n ],\n \"functions\": [\n {\n \"name\": \"safe_path\",\n \"signature\": \"def safe_path(p: str)\",\n \"startLine\": 130\n },\n {\n \"name\": \"run_bash\",\n \"signature\": \"def run_bash(command: str)\",\n \"startLine\": 136\n },\n {\n \"name\": \"run_read\",\n \"signature\": \"def run_read(path: str, limit: int = None)\",\n \"startLine\": 148\n },\n {\n \"name\": \"run_write\",\n \"signature\": \"def run_write(path: str, content: str)\",\n \"startLine\": 157\n },\n {\n \"name\": \"run_edit\",\n \"signature\": \"def run_edit(path: str, old_text: str, new_text: str)\",\n \"startLine\": 166\n },\n {\n \"name\": \"agent_loop\",\n \"signature\": \"def agent_loop(messages: list)\",\n \"startLine\": 209\n }\n ],\n \"layer\": \"planning\",\n \"source\": \"#!/usr/bin/env python3\\n\\\"\\\"\\\"\\ns07_task_system.py - Tasks\\n\\nTasks persist as JSON files in .tasks/ so they survive context compression.\\nEach task has a dependency graph (blockedBy/blocks).\\n\\n .tasks/\\n task_1.json {\\\"id\\\":1, \\\"subject\\\":\\\"...\\\", \\\"status\\\":\\\"completed\\\", ...}\\n task_2.json {\\\"id\\\":2, \\\"blockedBy\\\":[1], \\\"status\\\":\\\"pending\\\", ...}\\n task_3.json {\\\"id\\\":3, \\\"blockedBy\\\":[2], \\\"blocks\\\":[], ...}\\n\\n Dependency resolution:\\n +----------+ +----------+ +----------+\\n | task 1 | --> | task 2 | --> | task 3 |\\n | complete | | blocked | | blocked |\\n +----------+ +----------+ +----------+\\n | ^\\n +--- completing task 1 removes it from task 2's blockedBy\\n\\nKey insight: \\\"State that survives compression -- because it's outside the conversation.\\\"\\n\\\"\\\"\\\"\\n\\nimport json\\nimport os\\nimport subprocess\\nfrom pathlib import Path\\n\\nfrom anthropic import Anthropic\\nfrom dotenv import load_dotenv\\n\\nload_dotenv(override=True)\\n\\nif os.getenv(\\\"ANTHROPIC_BASE_URL\\\"):\\n os.environ.pop(\\\"ANTHROPIC_AUTH_TOKEN\\\", None)\\n\\nWORKDIR = Path.cwd()\\nclient = Anthropic(base_url=os.getenv(\\\"ANTHROPIC_BASE_URL\\\"))\\nMODEL = os.environ[\\\"MODEL_ID\\\"]\\nTASKS_DIR = WORKDIR / \\\".tasks\\\"\\n\\nSYSTEM = f\\\"You are a coding agent at {WORKDIR}. Use task tools to plan and track work.\\\"\\n\\n\\n# -- TaskManager: CRUD with dependency graph, persisted as JSON files --\\nclass TaskManager:\\n def __init__(self, tasks_dir: Path):\\n self.dir = tasks_dir\\n self.dir.mkdir(exist_ok=True)\\n self._next_id = self._max_id() + 1\\n\\n def _max_id(self) -> int:\\n ids = [int(f.stem.split(\\\"_\\\")[1]) for f in self.dir.glob(\\\"task_*.json\\\")]\\n return max(ids) if ids else 0\\n\\n def _load(self, task_id: int) -> dict:\\n path = self.dir / f\\\"task_{task_id}.json\\\"\\n if not path.exists():\\n raise ValueError(f\\\"Task {task_id} not found\\\")\\n return json.loads(path.read_text())\\n\\n def _save(self, task: dict):\\n path = self.dir / f\\\"task_{task['id']}.json\\\"\\n path.write_text(json.dumps(task, indent=2))\\n\\n def create(self, subject: str, description: str = \\\"\\\") -> str:\\n task = {\\n \\\"id\\\": self._next_id, \\\"subject\\\": subject, \\\"description\\\": description,\\n \\\"status\\\": \\\"pending\\\", \\\"blockedBy\\\": [], \\\"blocks\\\": [], \\\"owner\\\": \\\"\\\",\\n }\\n self._save(task)\\n self._next_id += 1\\n return json.dumps(task, indent=2)\\n\\n def get(self, task_id: int) -> str:\\n return json.dumps(self._load(task_id), indent=2)\\n\\n def update(self, task_id: int, status: str = None,\\n add_blocked_by: list = None, add_blocks: list = None) -> str:\\n task = self._load(task_id)\\n if status:\\n if status not in (\\\"pending\\\", \\\"in_progress\\\", \\\"completed\\\"):\\n raise ValueError(f\\\"Invalid status: {status}\\\")\\n task[\\\"status\\\"] = status\\n # When a task is completed, remove it from all other tasks' blockedBy\\n if status == \\\"completed\\\":\\n self._clear_dependency(task_id)\\n if add_blocked_by:\\n task[\\\"blockedBy\\\"] = list(set(task[\\\"blockedBy\\\"] + add_blocked_by))\\n if add_blocks:\\n task[\\\"blocks\\\"] = list(set(task[\\\"blocks\\\"] + add_blocks))\\n # Bidirectional: also update the blocked tasks' blockedBy lists\\n for blocked_id in add_blocks:\\n try:\\n blocked = self._load(blocked_id)\\n if task_id not in blocked[\\\"blockedBy\\\"]:\\n blocked[\\\"blockedBy\\\"].append(task_id)\\n self._save(blocked)\\n except ValueError:\\n pass\\n self._save(task)\\n return json.dumps(task, indent=2)\\n\\n def _clear_dependency(self, completed_id: int):\\n \\\"\\\"\\\"Remove completed_id from all other tasks' blockedBy lists.\\\"\\\"\\\"\\n for f in self.dir.glob(\\\"task_*.json\\\"):\\n task = json.loads(f.read_text())\\n if completed_id in task.get(\\\"blockedBy\\\", []):\\n task[\\\"blockedBy\\\"].remove(completed_id)\\n self._save(task)\\n\\n def list_all(self) -> str:\\n tasks = []\\n for f in sorted(self.dir.glob(\\\"task_*.json\\\")):\\n tasks.append(json.loads(f.read_text()))\\n if not tasks:\\n return \\\"No tasks.\\\"\\n lines = []\\n for t in tasks:\\n marker = {\\\"pending\\\": \\\"[ ]\\\", \\\"in_progress\\\": \\\"[>]\\\", \\\"completed\\\": \\\"[x]\\\"}.get(t[\\\"status\\\"], \\\"[?]\\\")\\n blocked = f\\\" (blocked by: {t['blockedBy']})\\\" if t.get(\\\"blockedBy\\\") else \\\"\\\"\\n lines.append(f\\\"{marker} #{t['id']}: {t['subject']}{blocked}\\\")\\n return \\\"\\\\n\\\".join(lines)\\n\\n\\nTASKS = TaskManager(TASKS_DIR)\\n\\n\\n# -- Base tool implementations --\\ndef safe_path(p: str) -> Path:\\n path = (WORKDIR / p).resolve()\\n if not path.is_relative_to(WORKDIR):\\n raise ValueError(f\\\"Path escapes workspace: {p}\\\")\\n return path\\n\\ndef run_bash(command: str) -> str:\\n dangerous = [\\\"rm -rf /\\\", \\\"sudo\\\", \\\"shutdown\\\", \\\"reboot\\\", \\\"> /dev/\\\"]\\n if any(d in command for d in dangerous):\\n return \\\"Error: Dangerous command blocked\\\"\\n try:\\n r = subprocess.run(command, shell=True, cwd=WORKDIR,\\n capture_output=True, text=True, timeout=120)\\n out = (r.stdout + r.stderr).strip()\\n return out[:50000] if out else \\\"(no output)\\\"\\n except subprocess.TimeoutExpired:\\n return \\\"Error: Timeout (120s)\\\"\\n\\ndef run_read(path: str, limit: int = None) -> str:\\n try:\\n lines = safe_path(path).read_text().splitlines()\\n if limit and limit < len(lines):\\n lines = lines[:limit] + [f\\\"... ({len(lines) - limit} more)\\\"]\\n return \\\"\\\\n\\\".join(lines)[:50000]\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\ndef run_write(path: str, content: str) -> str:\\n try:\\n fp = safe_path(path)\\n fp.parent.mkdir(parents=True, exist_ok=True)\\n fp.write_text(content)\\n return f\\\"Wrote {len(content)} bytes\\\"\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\ndef run_edit(path: str, old_text: str, new_text: str) -> str:\\n try:\\n fp = safe_path(path)\\n c = fp.read_text()\\n if old_text not in c:\\n return f\\\"Error: Text not found in {path}\\\"\\n fp.write_text(c.replace(old_text, new_text, 1))\\n return f\\\"Edited {path}\\\"\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\n\\nTOOL_HANDLERS = {\\n \\\"bash\\\": lambda **kw: run_bash(kw[\\\"command\\\"]),\\n \\\"read_file\\\": lambda **kw: run_read(kw[\\\"path\\\"], kw.get(\\\"limit\\\")),\\n \\\"write_file\\\": lambda **kw: run_write(kw[\\\"path\\\"], kw[\\\"content\\\"]),\\n \\\"edit_file\\\": lambda **kw: run_edit(kw[\\\"path\\\"], kw[\\\"old_text\\\"], kw[\\\"new_text\\\"]),\\n \\\"task_create\\\": lambda **kw: TASKS.create(kw[\\\"subject\\\"], kw.get(\\\"description\\\", \\\"\\\")),\\n \\\"task_update\\\": lambda **kw: TASKS.update(kw[\\\"task_id\\\"], kw.get(\\\"status\\\"), kw.get(\\\"addBlockedBy\\\"), kw.get(\\\"addBlocks\\\")),\\n \\\"task_list\\\": lambda **kw: TASKS.list_all(),\\n \\\"task_get\\\": lambda **kw: TASKS.get(kw[\\\"task_id\\\"]),\\n}\\n\\nTOOLS = [\\n {\\\"name\\\": \\\"bash\\\", \\\"description\\\": \\\"Run a shell command.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"command\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"command\\\"]}},\\n {\\\"name\\\": \\\"read_file\\\", \\\"description\\\": \\\"Read file contents.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"limit\\\": {\\\"type\\\": \\\"integer\\\"}}, \\\"required\\\": [\\\"path\\\"]}},\\n {\\\"name\\\": \\\"write_file\\\", \\\"description\\\": \\\"Write content to file.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"content\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\", \\\"content\\\"]}},\\n {\\\"name\\\": \\\"edit_file\\\", \\\"description\\\": \\\"Replace exact text in file.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"old_text\\\": {\\\"type\\\": \\\"string\\\"}, \\\"new_text\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\", \\\"old_text\\\", \\\"new_text\\\"]}},\\n {\\\"name\\\": \\\"task_create\\\", \\\"description\\\": \\\"Create a new task.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"subject\\\": {\\\"type\\\": \\\"string\\\"}, \\\"description\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"subject\\\"]}},\\n {\\\"name\\\": \\\"task_update\\\", \\\"description\\\": \\\"Update a task's status or dependencies.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"task_id\\\": {\\\"type\\\": \\\"integer\\\"}, \\\"status\\\": {\\\"type\\\": \\\"string\\\", \\\"enum\\\": [\\\"pending\\\", \\\"in_progress\\\", \\\"completed\\\"]}, \\\"addBlockedBy\\\": {\\\"type\\\": \\\"array\\\", \\\"items\\\": {\\\"type\\\": \\\"integer\\\"}}, \\\"addBlocks\\\": {\\\"type\\\": \\\"array\\\", \\\"items\\\": {\\\"type\\\": \\\"integer\\\"}}}, \\\"required\\\": [\\\"task_id\\\"]}},\\n {\\\"name\\\": \\\"task_list\\\", \\\"description\\\": \\\"List all tasks with status summary.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {}}},\\n {\\\"name\\\": \\\"task_get\\\", \\\"description\\\": \\\"Get full details of a task by ID.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"task_id\\\": {\\\"type\\\": \\\"integer\\\"}}, \\\"required\\\": [\\\"task_id\\\"]}},\\n]\\n\\n\\ndef agent_loop(messages: list):\\n while True:\\n response = client.messages.create(\\n model=MODEL, system=SYSTEM, messages=messages,\\n tools=TOOLS, max_tokens=8000,\\n )\\n messages.append({\\\"role\\\": \\\"assistant\\\", \\\"content\\\": response.content})\\n if response.stop_reason != \\\"tool_use\\\":\\n return\\n results = []\\n for block in response.content:\\n if block.type == \\\"tool_use\\\":\\n handler = TOOL_HANDLERS.get(block.name)\\n try:\\n output = handler(**block.input) if handler else f\\\"Unknown tool: {block.name}\\\"\\n except Exception as e:\\n output = f\\\"Error: {e}\\\"\\n print(f\\\"> {block.name}: {str(output)[:200]}\\\")\\n results.append({\\\"type\\\": \\\"tool_result\\\", \\\"tool_use_id\\\": block.id, \\\"content\\\": str(output)})\\n messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": results})\\n\\n\\nif __name__ == \\\"__main__\\\":\\n history = []\\n while True:\\n try:\\n query = input(\\\"\\\\033[36ms07 >> \\\\033[0m\\\")\\n except (EOFError, KeyboardInterrupt):\\n break\\n if query.strip().lower() in (\\\"q\\\", \\\"exit\\\", \\\"\\\"):\\n break\\n history.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": query})\\n agent_loop(history)\\n print()\\n\"\n },\n {\n \"id\": \"s08\",\n \"filename\": \"s08_background_tasks.py\",\n \"title\": \"Background Tasks\",\n \"subtitle\": \"Background Threads + Notifications\",\n \"loc\": 193,\n \"tools\": [\n \"bash\",\n \"read_file\",\n \"write_file\",\n \"edit_file\",\n \"background_run\",\n \"check_background\"\n ],\n \"newTools\": [\n \"background_run\",\n \"check_background\"\n ],\n \"coreAddition\": \"BackgroundManager + notification queue\",\n \"keyInsight\": \"Run slow operations in the background; the agent keeps thinking ahead\",\n \"classes\": [\n {\n \"name\": \"BackgroundManager\",\n \"startLine\": 49,\n \"endLine\": 109\n }\n ],\n \"functions\": [\n {\n \"name\": \"safe_path\",\n \"signature\": \"def safe_path(p: str)\",\n \"startLine\": 114\n },\n {\n \"name\": \"run_bash\",\n \"signature\": \"def run_bash(command: str)\",\n \"startLine\": 120\n },\n {\n \"name\": \"run_read\",\n \"signature\": \"def run_read(path: str, limit: int = None)\",\n \"startLine\": 132\n },\n {\n \"name\": \"run_write\",\n \"signature\": \"def run_write(path: str, content: str)\",\n \"startLine\": 141\n },\n {\n \"name\": \"run_edit\",\n \"signature\": \"def run_edit(path: str, old_text: str, new_text: str)\",\n \"startLine\": 150\n },\n {\n \"name\": \"agent_loop\",\n \"signature\": \"def agent_loop(messages: list)\",\n \"startLine\": 187\n }\n ],\n \"layer\": \"concurrency\",\n \"source\": \"#!/usr/bin/env python3\\n\\\"\\\"\\\"\\ns08_background_tasks.py - Background Tasks\\n\\nRun commands in background threads. A notification queue is drained\\nbefore each LLM call to deliver results.\\n\\n Main thread Background thread\\n +-----------------+ +-----------------+\\n | agent loop | | task executes |\\n | ... | | ... |\\n | [LLM call] <---+------- | enqueue(result) |\\n | ^drain queue | +-----------------+\\n +-----------------+\\n\\n Timeline:\\n Agent ----[spawn A]----[spawn B]----[other work]----\\n | |\\n v v\\n [A runs] [B runs] (parallel)\\n | |\\n +-- notification queue --> [results injected]\\n\\nKey insight: \\\"Fire and forget -- the agent doesn't block while the command runs.\\\"\\n\\\"\\\"\\\"\\n\\nimport os\\nimport subprocess\\nimport threading\\nimport uuid\\nfrom pathlib import Path\\n\\nfrom anthropic import Anthropic\\nfrom dotenv import load_dotenv\\n\\nload_dotenv(override=True)\\n\\nif os.getenv(\\\"ANTHROPIC_BASE_URL\\\"):\\n os.environ.pop(\\\"ANTHROPIC_AUTH_TOKEN\\\", None)\\n\\nWORKDIR = Path.cwd()\\nclient = Anthropic(base_url=os.getenv(\\\"ANTHROPIC_BASE_URL\\\"))\\nMODEL = os.environ[\\\"MODEL_ID\\\"]\\n\\nSYSTEM = f\\\"You are a coding agent at {WORKDIR}. Use background_run for long-running commands.\\\"\\n\\n\\n# -- BackgroundManager: threaded execution + notification queue --\\nclass BackgroundManager:\\n def __init__(self):\\n self.tasks = {} # task_id -> {status, result, command}\\n self._notification_queue = [] # completed task results\\n self._lock = threading.Lock()\\n\\n def run(self, command: str) -> str:\\n \\\"\\\"\\\"Start a background thread, return task_id immediately.\\\"\\\"\\\"\\n task_id = str(uuid.uuid4())[:8]\\n self.tasks[task_id] = {\\\"status\\\": \\\"running\\\", \\\"result\\\": None, \\\"command\\\": command}\\n thread = threading.Thread(\\n target=self._execute, args=(task_id, command), daemon=True\\n )\\n thread.start()\\n return f\\\"Background task {task_id} started: {command[:80]}\\\"\\n\\n def _execute(self, task_id: str, command: str):\\n \\\"\\\"\\\"Thread target: run subprocess, capture output, push to queue.\\\"\\\"\\\"\\n try:\\n r = subprocess.run(\\n command, shell=True, cwd=WORKDIR,\\n capture_output=True, text=True, timeout=300\\n )\\n output = (r.stdout + r.stderr).strip()[:50000]\\n status = \\\"completed\\\"\\n except subprocess.TimeoutExpired:\\n output = \\\"Error: Timeout (300s)\\\"\\n status = \\\"timeout\\\"\\n except Exception as e:\\n output = f\\\"Error: {e}\\\"\\n status = \\\"error\\\"\\n self.tasks[task_id][\\\"status\\\"] = status\\n self.tasks[task_id][\\\"result\\\"] = output or \\\"(no output)\\\"\\n with self._lock:\\n self._notification_queue.append({\\n \\\"task_id\\\": task_id,\\n \\\"status\\\": status,\\n \\\"command\\\": command[:80],\\n \\\"result\\\": (output or \\\"(no output)\\\")[:500],\\n })\\n\\n def check(self, task_id: str = None) -> str:\\n \\\"\\\"\\\"Check status of one task or list all.\\\"\\\"\\\"\\n if task_id:\\n t = self.tasks.get(task_id)\\n if not t:\\n return f\\\"Error: Unknown task {task_id}\\\"\\n return f\\\"[{t['status']}] {t['command'][:60]}\\\\n{t.get('result') or '(running)'}\\\"\\n lines = []\\n for tid, t in self.tasks.items():\\n lines.append(f\\\"{tid}: [{t['status']}] {t['command'][:60]}\\\")\\n return \\\"\\\\n\\\".join(lines) if lines else \\\"No background tasks.\\\"\\n\\n def drain_notifications(self) -> list:\\n \\\"\\\"\\\"Return and clear all pending completion notifications.\\\"\\\"\\\"\\n with self._lock:\\n notifs = list(self._notification_queue)\\n self._notification_queue.clear()\\n return notifs\\n\\n\\nBG = BackgroundManager()\\n\\n\\n# -- Tool implementations --\\ndef safe_path(p: str) -> Path:\\n path = (WORKDIR / p).resolve()\\n if not path.is_relative_to(WORKDIR):\\n raise ValueError(f\\\"Path escapes workspace: {p}\\\")\\n return path\\n\\ndef run_bash(command: str) -> str:\\n dangerous = [\\\"rm -rf /\\\", \\\"sudo\\\", \\\"shutdown\\\", \\\"reboot\\\", \\\"> /dev/\\\"]\\n if any(d in command for d in dangerous):\\n return \\\"Error: Dangerous command blocked\\\"\\n try:\\n r = subprocess.run(command, shell=True, cwd=WORKDIR,\\n capture_output=True, text=True, timeout=120)\\n out = (r.stdout + r.stderr).strip()\\n return out[:50000] if out else \\\"(no output)\\\"\\n except subprocess.TimeoutExpired:\\n return \\\"Error: Timeout (120s)\\\"\\n\\ndef run_read(path: str, limit: int = None) -> str:\\n try:\\n lines = safe_path(path).read_text().splitlines()\\n if limit and limit < len(lines):\\n lines = lines[:limit] + [f\\\"... ({len(lines) - limit} more)\\\"]\\n return \\\"\\\\n\\\".join(lines)[:50000]\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\ndef run_write(path: str, content: str) -> str:\\n try:\\n fp = safe_path(path)\\n fp.parent.mkdir(parents=True, exist_ok=True)\\n fp.write_text(content)\\n return f\\\"Wrote {len(content)} bytes\\\"\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\ndef run_edit(path: str, old_text: str, new_text: str) -> str:\\n try:\\n fp = safe_path(path)\\n c = fp.read_text()\\n if old_text not in c:\\n return f\\\"Error: Text not found in {path}\\\"\\n fp.write_text(c.replace(old_text, new_text, 1))\\n return f\\\"Edited {path}\\\"\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\n\\nTOOL_HANDLERS = {\\n \\\"bash\\\": lambda **kw: run_bash(kw[\\\"command\\\"]),\\n \\\"read_file\\\": lambda **kw: run_read(kw[\\\"path\\\"], kw.get(\\\"limit\\\")),\\n \\\"write_file\\\": lambda **kw: run_write(kw[\\\"path\\\"], kw[\\\"content\\\"]),\\n \\\"edit_file\\\": lambda **kw: run_edit(kw[\\\"path\\\"], kw[\\\"old_text\\\"], kw[\\\"new_text\\\"]),\\n \\\"background_run\\\": lambda **kw: BG.run(kw[\\\"command\\\"]),\\n \\\"check_background\\\": lambda **kw: BG.check(kw.get(\\\"task_id\\\")),\\n}\\n\\nTOOLS = [\\n {\\\"name\\\": \\\"bash\\\", \\\"description\\\": \\\"Run a shell command (blocking).\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"command\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"command\\\"]}},\\n {\\\"name\\\": \\\"read_file\\\", \\\"description\\\": \\\"Read file contents.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"limit\\\": {\\\"type\\\": \\\"integer\\\"}}, \\\"required\\\": [\\\"path\\\"]}},\\n {\\\"name\\\": \\\"write_file\\\", \\\"description\\\": \\\"Write content to file.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"content\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\", \\\"content\\\"]}},\\n {\\\"name\\\": \\\"edit_file\\\", \\\"description\\\": \\\"Replace exact text in file.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"old_text\\\": {\\\"type\\\": \\\"string\\\"}, \\\"new_text\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\", \\\"old_text\\\", \\\"new_text\\\"]}},\\n {\\\"name\\\": \\\"background_run\\\", \\\"description\\\": \\\"Run command in background thread. Returns task_id immediately.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"command\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"command\\\"]}},\\n {\\\"name\\\": \\\"check_background\\\", \\\"description\\\": \\\"Check background task status. Omit task_id to list all.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"task_id\\\": {\\\"type\\\": \\\"string\\\"}}}},\\n]\\n\\n\\ndef agent_loop(messages: list):\\n while True:\\n # Drain background notifications and inject as system message before LLM call\\n notifs = BG.drain_notifications()\\n if notifs and messages:\\n notif_text = \\\"\\\\n\\\".join(\\n f\\\"[bg:{n['task_id']}] {n['status']}: {n['result']}\\\" for n in notifs\\n )\\n messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": f\\\"\\\\n{notif_text}\\\\n \\\"})\\n messages.append({\\\"role\\\": \\\"assistant\\\", \\\"content\\\": \\\"Noted background results.\\\"})\\n response = client.messages.create(\\n model=MODEL, system=SYSTEM, messages=messages,\\n tools=TOOLS, max_tokens=8000,\\n )\\n messages.append({\\\"role\\\": \\\"assistant\\\", \\\"content\\\": response.content})\\n if response.stop_reason != \\\"tool_use\\\":\\n return\\n results = []\\n for block in response.content:\\n if block.type == \\\"tool_use\\\":\\n handler = TOOL_HANDLERS.get(block.name)\\n try:\\n output = handler(**block.input) if handler else f\\\"Unknown tool: {block.name}\\\"\\n except Exception as e:\\n output = f\\\"Error: {e}\\\"\\n print(f\\\"> {block.name}: {str(output)[:200]}\\\")\\n results.append({\\\"type\\\": \\\"tool_result\\\", \\\"tool_use_id\\\": block.id, \\\"content\\\": str(output)})\\n messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": results})\\n\\n\\nif __name__ == \\\"__main__\\\":\\n history = []\\n while True:\\n try:\\n query = input(\\\"\\\\033[36ms08 >> \\\\033[0m\\\")\\n except (EOFError, KeyboardInterrupt):\\n break\\n if query.strip().lower() in (\\\"q\\\", \\\"exit\\\", \\\"\\\"):\\n break\\n history.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": query})\\n agent_loop(history)\\n print()\\n\"\n },\n {\n \"id\": \"s09\",\n \"filename\": \"s09_agent_teams.py\",\n \"title\": \"Agent Teams\",\n \"subtitle\": \"Teammates + Mailboxes\",\n \"loc\": 343,\n \"tools\": [\n \"alice\",\n \"bash\",\n \"read_file\",\n \"write_file\",\n \"edit_file\",\n \"send_message\",\n \"read_inbox\",\n \"spawn_teammate\",\n \"list_teammates\",\n \"broadcast\"\n ],\n \"newTools\": [\n \"alice\",\n \"send_message\",\n \"read_inbox\",\n \"spawn_teammate\",\n \"list_teammates\",\n \"broadcast\"\n ],\n \"coreAddition\": \"TeammateManager + file-based mailbox\",\n \"keyInsight\": \"When one agent can't finish, delegate to persistent teammates via async mailboxes\",\n \"classes\": [\n {\n \"name\": \"MessageBus\",\n \"startLine\": 77,\n \"endLine\": 118\n },\n {\n \"name\": \"TeammateManager\",\n \"startLine\": 123,\n \"endLine\": 249\n }\n ],\n \"functions\": [\n {\n \"name\": \"_safe_path\",\n \"signature\": \"def _safe_path(p: str)\",\n \"startLine\": 254\n },\n {\n \"name\": \"_run_bash\",\n \"signature\": \"def _run_bash(command: str)\",\n \"startLine\": 261\n },\n {\n \"name\": \"_run_read\",\n \"signature\": \"def _run_read(path: str, limit: int = None)\",\n \"startLine\": 276\n },\n {\n \"name\": \"_run_write\",\n \"signature\": \"def _run_write(path: str, content: str)\",\n \"startLine\": 286\n },\n {\n \"name\": \"_run_edit\",\n \"signature\": \"def _run_edit(path: str, old_text: str, new_text: str)\",\n \"startLine\": 296\n },\n {\n \"name\": \"agent_loop\",\n \"signature\": \"def agent_loop(messages: list)\",\n \"startLine\": 344\n }\n ],\n \"layer\": \"collaboration\",\n \"source\": \"#!/usr/bin/env python3\\n\\\"\\\"\\\"\\ns09_agent_teams.py - Agent Teams\\n\\nPersistent named agents with file-based JSONL inboxes. Each teammate runs\\nits own agent loop in a separate thread. Communication via append-only inboxes.\\n\\n Subagent (s04): spawn -> execute -> return summary -> destroyed\\n Teammate (s09): spawn -> work -> idle -> work -> ... -> shutdown\\n\\n .team/config.json .team/inbox/\\n +----------------------------+ +------------------+\\n | {\\\"team_name\\\": \\\"default\\\", | | alice.jsonl |\\n | \\\"members\\\": [ | | bob.jsonl |\\n | {\\\"name\\\":\\\"alice\\\", | | lead.jsonl |\\n | \\\"role\\\":\\\"coder\\\", | +------------------+\\n | \\\"status\\\":\\\"idle\\\"} |\\n | ]} | send_message(\\\"alice\\\", \\\"fix bug\\\"):\\n +----------------------------+ open(\\\"alice.jsonl\\\", \\\"a\\\").write(msg)\\n\\n read_inbox(\\\"alice\\\"):\\n spawn_teammate(\\\"alice\\\",\\\"coder\\\",...) msgs = [json.loads(l) for l in ...]\\n | open(\\\"alice.jsonl\\\", \\\"w\\\").close()\\n v return msgs # drain\\n Thread: alice Thread: bob\\n +------------------+ +------------------+\\n | agent_loop | | agent_loop |\\n | status: working | | status: idle |\\n | ... runs tools | | ... waits ... |\\n | status -> idle | | |\\n +------------------+ +------------------+\\n\\n 5 message types (all declared, not all handled here):\\n +-------------------------+-----------------------------------+\\n | message | Normal text message |\\n | broadcast | Sent to all teammates |\\n | shutdown_request | Request graceful shutdown (s10) |\\n | shutdown_response | Approve/reject shutdown (s10) |\\n | plan_approval_response | Approve/reject plan (s10) |\\n +-------------------------+-----------------------------------+\\n\\nKey insight: \\\"Teammates that can talk to each other.\\\"\\n\\\"\\\"\\\"\\n\\nimport json\\nimport os\\nimport subprocess\\nimport threading\\nimport time\\nfrom pathlib import Path\\n\\nfrom anthropic import Anthropic\\nfrom dotenv import load_dotenv\\n\\nload_dotenv(override=True)\\nif os.getenv(\\\"ANTHROPIC_BASE_URL\\\"):\\n os.environ.pop(\\\"ANTHROPIC_AUTH_TOKEN\\\", None)\\n\\nWORKDIR = Path.cwd()\\nclient = Anthropic(base_url=os.getenv(\\\"ANTHROPIC_BASE_URL\\\"))\\nMODEL = os.environ[\\\"MODEL_ID\\\"]\\nTEAM_DIR = WORKDIR / \\\".team\\\"\\nINBOX_DIR = TEAM_DIR / \\\"inbox\\\"\\n\\nSYSTEM = f\\\"You are a team lead at {WORKDIR}. Spawn teammates and communicate via inboxes.\\\"\\n\\nVALID_MSG_TYPES = {\\n \\\"message\\\",\\n \\\"broadcast\\\",\\n \\\"shutdown_request\\\",\\n \\\"shutdown_response\\\",\\n \\\"plan_approval_response\\\",\\n}\\n\\n\\n# -- MessageBus: JSONL inbox per teammate --\\nclass MessageBus:\\n def __init__(self, inbox_dir: Path):\\n self.dir = inbox_dir\\n self.dir.mkdir(parents=True, exist_ok=True)\\n\\n def send(self, sender: str, to: str, content: str,\\n msg_type: str = \\\"message\\\", extra: dict = None) -> str:\\n if msg_type not in VALID_MSG_TYPES:\\n return f\\\"Error: Invalid type '{msg_type}'. Valid: {VALID_MSG_TYPES}\\\"\\n msg = {\\n \\\"type\\\": msg_type,\\n \\\"from\\\": sender,\\n \\\"content\\\": content,\\n \\\"timestamp\\\": time.time(),\\n }\\n if extra:\\n msg.update(extra)\\n inbox_path = self.dir / f\\\"{to}.jsonl\\\"\\n with open(inbox_path, \\\"a\\\") as f:\\n f.write(json.dumps(msg) + \\\"\\\\n\\\")\\n return f\\\"Sent {msg_type} to {to}\\\"\\n\\n def read_inbox(self, name: str) -> list:\\n inbox_path = self.dir / f\\\"{name}.jsonl\\\"\\n if not inbox_path.exists():\\n return []\\n messages = []\\n for line in inbox_path.read_text().strip().splitlines():\\n if line:\\n messages.append(json.loads(line))\\n inbox_path.write_text(\\\"\\\")\\n return messages\\n\\n def broadcast(self, sender: str, content: str, teammates: list) -> str:\\n count = 0\\n for name in teammates:\\n if name != sender:\\n self.send(sender, name, content, \\\"broadcast\\\")\\n count += 1\\n return f\\\"Broadcast to {count} teammates\\\"\\n\\n\\nBUS = MessageBus(INBOX_DIR)\\n\\n\\n# -- TeammateManager: persistent named agents with config.json --\\nclass TeammateManager:\\n def __init__(self, team_dir: Path):\\n self.dir = team_dir\\n self.dir.mkdir(exist_ok=True)\\n self.config_path = self.dir / \\\"config.json\\\"\\n self.config = self._load_config()\\n self.threads = {}\\n\\n def _load_config(self) -> dict:\\n if self.config_path.exists():\\n return json.loads(self.config_path.read_text())\\n return {\\\"team_name\\\": \\\"default\\\", \\\"members\\\": []}\\n\\n def _save_config(self):\\n self.config_path.write_text(json.dumps(self.config, indent=2))\\n\\n def _find_member(self, name: str) -> dict:\\n for m in self.config[\\\"members\\\"]:\\n if m[\\\"name\\\"] == name:\\n return m\\n return None\\n\\n def spawn(self, name: str, role: str, prompt: str) -> str:\\n member = self._find_member(name)\\n if member:\\n if member[\\\"status\\\"] not in (\\\"idle\\\", \\\"shutdown\\\"):\\n return f\\\"Error: '{name}' is currently {member['status']}\\\"\\n member[\\\"status\\\"] = \\\"working\\\"\\n member[\\\"role\\\"] = role\\n else:\\n member = {\\\"name\\\": name, \\\"role\\\": role, \\\"status\\\": \\\"working\\\"}\\n self.config[\\\"members\\\"].append(member)\\n self._save_config()\\n thread = threading.Thread(\\n target=self._teammate_loop,\\n args=(name, role, prompt),\\n daemon=True,\\n )\\n self.threads[name] = thread\\n thread.start()\\n return f\\\"Spawned '{name}' (role: {role})\\\"\\n\\n def _teammate_loop(self, name: str, role: str, prompt: str):\\n sys_prompt = (\\n f\\\"You are '{name}', role: {role}, at {WORKDIR}. \\\"\\n f\\\"Use send_message to communicate. Complete your task.\\\"\\n )\\n messages = [{\\\"role\\\": \\\"user\\\", \\\"content\\\": prompt}]\\n tools = self._teammate_tools()\\n for _ in range(50):\\n inbox = BUS.read_inbox(name)\\n for msg in inbox:\\n messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": json.dumps(msg)})\\n try:\\n response = client.messages.create(\\n model=MODEL,\\n system=sys_prompt,\\n messages=messages,\\n tools=tools,\\n max_tokens=8000,\\n )\\n except Exception:\\n break\\n messages.append({\\\"role\\\": \\\"assistant\\\", \\\"content\\\": response.content})\\n if response.stop_reason != \\\"tool_use\\\":\\n break\\n results = []\\n for block in response.content:\\n if block.type == \\\"tool_use\\\":\\n output = self._exec(name, block.name, block.input)\\n print(f\\\" [{name}] {block.name}: {str(output)[:120]}\\\")\\n results.append({\\n \\\"type\\\": \\\"tool_result\\\",\\n \\\"tool_use_id\\\": block.id,\\n \\\"content\\\": str(output),\\n })\\n messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": results})\\n member = self._find_member(name)\\n if member and member[\\\"status\\\"] != \\\"shutdown\\\":\\n member[\\\"status\\\"] = \\\"idle\\\"\\n self._save_config()\\n\\n def _exec(self, sender: str, tool_name: str, args: dict) -> str:\\n # these base tools are unchanged from s02\\n if tool_name == \\\"bash\\\":\\n return _run_bash(args[\\\"command\\\"])\\n if tool_name == \\\"read_file\\\":\\n return _run_read(args[\\\"path\\\"])\\n if tool_name == \\\"write_file\\\":\\n return _run_write(args[\\\"path\\\"], args[\\\"content\\\"])\\n if tool_name == \\\"edit_file\\\":\\n return _run_edit(args[\\\"path\\\"], args[\\\"old_text\\\"], args[\\\"new_text\\\"])\\n if tool_name == \\\"send_message\\\":\\n return BUS.send(sender, args[\\\"to\\\"], args[\\\"content\\\"], args.get(\\\"msg_type\\\", \\\"message\\\"))\\n if tool_name == \\\"read_inbox\\\":\\n return json.dumps(BUS.read_inbox(sender), indent=2)\\n return f\\\"Unknown tool: {tool_name}\\\"\\n\\n def _teammate_tools(self) -> list:\\n # these base tools are unchanged from s02\\n return [\\n {\\\"name\\\": \\\"bash\\\", \\\"description\\\": \\\"Run a shell command.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"command\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"command\\\"]}},\\n {\\\"name\\\": \\\"read_file\\\", \\\"description\\\": \\\"Read file contents.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\"]}},\\n {\\\"name\\\": \\\"write_file\\\", \\\"description\\\": \\\"Write content to file.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"content\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\", \\\"content\\\"]}},\\n {\\\"name\\\": \\\"edit_file\\\", \\\"description\\\": \\\"Replace exact text in file.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"old_text\\\": {\\\"type\\\": \\\"string\\\"}, \\\"new_text\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\", \\\"old_text\\\", \\\"new_text\\\"]}},\\n {\\\"name\\\": \\\"send_message\\\", \\\"description\\\": \\\"Send message to a teammate.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"to\\\": {\\\"type\\\": \\\"string\\\"}, \\\"content\\\": {\\\"type\\\": \\\"string\\\"}, \\\"msg_type\\\": {\\\"type\\\": \\\"string\\\", \\\"enum\\\": list(VALID_MSG_TYPES)}}, \\\"required\\\": [\\\"to\\\", \\\"content\\\"]}},\\n {\\\"name\\\": \\\"read_inbox\\\", \\\"description\\\": \\\"Read and drain your inbox.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {}}},\\n ]\\n\\n def list_all(self) -> str:\\n if not self.config[\\\"members\\\"]:\\n return \\\"No teammates.\\\"\\n lines = [f\\\"Team: {self.config['team_name']}\\\"]\\n for m in self.config[\\\"members\\\"]:\\n lines.append(f\\\" {m['name']} ({m['role']}): {m['status']}\\\")\\n return \\\"\\\\n\\\".join(lines)\\n\\n def member_names(self) -> list:\\n return [m[\\\"name\\\"] for m in self.config[\\\"members\\\"]]\\n\\n\\nTEAM = TeammateManager(TEAM_DIR)\\n\\n\\n# -- Base tool implementations (these base tools are unchanged from s02) --\\ndef _safe_path(p: str) -> Path:\\n path = (WORKDIR / p).resolve()\\n if not path.is_relative_to(WORKDIR):\\n raise ValueError(f\\\"Path escapes workspace: {p}\\\")\\n return path\\n\\n\\ndef _run_bash(command: str) -> str:\\n dangerous = [\\\"rm -rf /\\\", \\\"sudo\\\", \\\"shutdown\\\", \\\"reboot\\\"]\\n if any(d in command for d in dangerous):\\n return \\\"Error: Dangerous command blocked\\\"\\n try:\\n r = subprocess.run(\\n command, shell=True, cwd=WORKDIR,\\n capture_output=True, text=True, timeout=120,\\n )\\n out = (r.stdout + r.stderr).strip()\\n return out[:50000] if out else \\\"(no output)\\\"\\n except subprocess.TimeoutExpired:\\n return \\\"Error: Timeout (120s)\\\"\\n\\n\\ndef _run_read(path: str, limit: int = None) -> str:\\n try:\\n lines = _safe_path(path).read_text().splitlines()\\n if limit and limit < len(lines):\\n lines = lines[:limit] + [f\\\"... ({len(lines) - limit} more)\\\"]\\n return \\\"\\\\n\\\".join(lines)[:50000]\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\n\\ndef _run_write(path: str, content: str) -> str:\\n try:\\n fp = _safe_path(path)\\n fp.parent.mkdir(parents=True, exist_ok=True)\\n fp.write_text(content)\\n return f\\\"Wrote {len(content)} bytes\\\"\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\n\\ndef _run_edit(path: str, old_text: str, new_text: str) -> str:\\n try:\\n fp = _safe_path(path)\\n c = fp.read_text()\\n if old_text not in c:\\n return f\\\"Error: Text not found in {path}\\\"\\n fp.write_text(c.replace(old_text, new_text, 1))\\n return f\\\"Edited {path}\\\"\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\n\\n# -- Lead tool dispatch (9 tools) --\\nTOOL_HANDLERS = {\\n \\\"bash\\\": lambda **kw: _run_bash(kw[\\\"command\\\"]),\\n \\\"read_file\\\": lambda **kw: _run_read(kw[\\\"path\\\"], kw.get(\\\"limit\\\")),\\n \\\"write_file\\\": lambda **kw: _run_write(kw[\\\"path\\\"], kw[\\\"content\\\"]),\\n \\\"edit_file\\\": lambda **kw: _run_edit(kw[\\\"path\\\"], kw[\\\"old_text\\\"], kw[\\\"new_text\\\"]),\\n \\\"spawn_teammate\\\": lambda **kw: TEAM.spawn(kw[\\\"name\\\"], kw[\\\"role\\\"], kw[\\\"prompt\\\"]),\\n \\\"list_teammates\\\": lambda **kw: TEAM.list_all(),\\n \\\"send_message\\\": lambda **kw: BUS.send(\\\"lead\\\", kw[\\\"to\\\"], kw[\\\"content\\\"], kw.get(\\\"msg_type\\\", \\\"message\\\")),\\n \\\"read_inbox\\\": lambda **kw: json.dumps(BUS.read_inbox(\\\"lead\\\"), indent=2),\\n \\\"broadcast\\\": lambda **kw: BUS.broadcast(\\\"lead\\\", kw[\\\"content\\\"], TEAM.member_names()),\\n}\\n\\n# these base tools are unchanged from s02\\nTOOLS = [\\n {\\\"name\\\": \\\"bash\\\", \\\"description\\\": \\\"Run a shell command.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"command\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"command\\\"]}},\\n {\\\"name\\\": \\\"read_file\\\", \\\"description\\\": \\\"Read file contents.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"limit\\\": {\\\"type\\\": \\\"integer\\\"}}, \\\"required\\\": [\\\"path\\\"]}},\\n {\\\"name\\\": \\\"write_file\\\", \\\"description\\\": \\\"Write content to file.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"content\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\", \\\"content\\\"]}},\\n {\\\"name\\\": \\\"edit_file\\\", \\\"description\\\": \\\"Replace exact text in file.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"old_text\\\": {\\\"type\\\": \\\"string\\\"}, \\\"new_text\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\", \\\"old_text\\\", \\\"new_text\\\"]}},\\n {\\\"name\\\": \\\"spawn_teammate\\\", \\\"description\\\": \\\"Spawn a persistent teammate that runs in its own thread.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"name\\\": {\\\"type\\\": \\\"string\\\"}, \\\"role\\\": {\\\"type\\\": \\\"string\\\"}, \\\"prompt\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"name\\\", \\\"role\\\", \\\"prompt\\\"]}},\\n {\\\"name\\\": \\\"list_teammates\\\", \\\"description\\\": \\\"List all teammates with name, role, status.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {}}},\\n {\\\"name\\\": \\\"send_message\\\", \\\"description\\\": \\\"Send a message to a teammate's inbox.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"to\\\": {\\\"type\\\": \\\"string\\\"}, \\\"content\\\": {\\\"type\\\": \\\"string\\\"}, \\\"msg_type\\\": {\\\"type\\\": \\\"string\\\", \\\"enum\\\": list(VALID_MSG_TYPES)}}, \\\"required\\\": [\\\"to\\\", \\\"content\\\"]}},\\n {\\\"name\\\": \\\"read_inbox\\\", \\\"description\\\": \\\"Read and drain the lead's inbox.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {}}},\\n {\\\"name\\\": \\\"broadcast\\\", \\\"description\\\": \\\"Send a message to all teammates.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"content\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"content\\\"]}},\\n]\\n\\n\\ndef agent_loop(messages: list):\\n while True:\\n inbox = BUS.read_inbox(\\\"lead\\\")\\n if inbox:\\n messages.append({\\n \\\"role\\\": \\\"user\\\",\\n \\\"content\\\": f\\\"{json.dumps(inbox, indent=2)} \\\",\\n })\\n messages.append({\\n \\\"role\\\": \\\"assistant\\\",\\n \\\"content\\\": \\\"Noted inbox messages.\\\",\\n })\\n response = client.messages.create(\\n model=MODEL,\\n system=SYSTEM,\\n messages=messages,\\n tools=TOOLS,\\n max_tokens=8000,\\n )\\n messages.append({\\\"role\\\": \\\"assistant\\\", \\\"content\\\": response.content})\\n if response.stop_reason != \\\"tool_use\\\":\\n return\\n results = []\\n for block in response.content:\\n if block.type == \\\"tool_use\\\":\\n handler = TOOL_HANDLERS.get(block.name)\\n try:\\n output = handler(**block.input) if handler else f\\\"Unknown tool: {block.name}\\\"\\n except Exception as e:\\n output = f\\\"Error: {e}\\\"\\n print(f\\\"> {block.name}: {str(output)[:200]}\\\")\\n results.append({\\n \\\"type\\\": \\\"tool_result\\\",\\n \\\"tool_use_id\\\": block.id,\\n \\\"content\\\": str(output),\\n })\\n messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": results})\\n\\n\\nif __name__ == \\\"__main__\\\":\\n history = []\\n while True:\\n try:\\n query = input(\\\"\\\\033[36ms09 >> \\\\033[0m\\\")\\n except (EOFError, KeyboardInterrupt):\\n break\\n if query.strip().lower() in (\\\"q\\\", \\\"exit\\\", \\\"\\\"):\\n break\\n if query.strip() == \\\"/team\\\":\\n print(TEAM.list_all())\\n continue\\n if query.strip() == \\\"/inbox\\\":\\n print(json.dumps(BUS.read_inbox(\\\"lead\\\"), indent=2))\\n continue\\n history.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": query})\\n agent_loop(history)\\n print()\\n\"\n },\n {\n \"id\": \"s10\",\n \"filename\": \"s10_team_protocols.py\",\n \"title\": \"Team Protocols\",\n \"subtitle\": \"Shared Communication Rules\",\n \"loc\": 414,\n \"tools\": [\n \"bash\",\n \"read_file\",\n \"write_file\",\n \"edit_file\",\n \"send_message\",\n \"read_inbox\",\n \"shutdown_response\",\n \"plan_approval\",\n \"spawn_teammate\",\n \"list_teammates\",\n \"broadcast\",\n \"shutdown_request\"\n ],\n \"newTools\": [\n \"shutdown_response\",\n \"plan_approval\",\n \"shutdown_request\"\n ],\n \"coreAddition\": \"request_id correlation for two protocols\",\n \"keyInsight\": \"One request-response pattern drives all team negotiation\",\n \"classes\": [\n {\n \"name\": \"MessageBus\",\n \"startLine\": 87,\n \"endLine\": 128\n },\n {\n \"name\": \"TeammateManager\",\n \"startLine\": 133,\n \"endLine\": 290\n }\n ],\n \"functions\": [\n {\n \"name\": \"_safe_path\",\n \"signature\": \"def _safe_path(p: str)\",\n \"startLine\": 295\n },\n {\n \"name\": \"_run_bash\",\n \"signature\": \"def _run_bash(command: str)\",\n \"startLine\": 302\n },\n {\n \"name\": \"_run_read\",\n \"signature\": \"def _run_read(path: str, limit: int = None)\",\n \"startLine\": 317\n },\n {\n \"name\": \"_run_write\",\n \"signature\": \"def _run_write(path: str, content: str)\",\n \"startLine\": 327\n },\n {\n \"name\": \"_run_edit\",\n \"signature\": \"def _run_edit(path: str, old_text: str, new_text: str)\",\n \"startLine\": 337\n },\n {\n \"name\": \"handle_shutdown_request\",\n \"signature\": \"def handle_shutdown_request(teammate: str)\",\n \"startLine\": 350\n },\n {\n \"name\": \"handle_plan_review\",\n \"signature\": \"def handle_plan_review(request_id: str, approve: bool, feedback: str = \\\"\\\")\",\n \"startLine\": 361\n },\n {\n \"name\": \"_check_shutdown_status\",\n \"signature\": \"def _check_shutdown_status(request_id: str)\",\n \"startLine\": 375\n },\n {\n \"name\": \"agent_loop\",\n \"signature\": \"def agent_loop(messages: list)\",\n \"startLine\": 425\n }\n ],\n \"layer\": \"collaboration\",\n \"source\": \"#!/usr/bin/env python3\\n\\\"\\\"\\\"\\ns10_team_protocols.py - Team Protocols\\n\\nShutdown protocol and plan approval protocol, both using the same\\nrequest_id correlation pattern. Builds on s09's team messaging.\\n\\n Shutdown FSM: pending -> approved | rejected\\n\\n Lead Teammate\\n +---------------------+ +---------------------+\\n | shutdown_request | | |\\n | { | -------> | receives request |\\n | request_id: abc | | decides: approve? |\\n | } | | |\\n +---------------------+ +---------------------+\\n |\\n +---------------------+ +-------v-------------+\\n | shutdown_response | <------- | shutdown_response |\\n | { | | { |\\n | request_id: abc | | request_id: abc |\\n | approve: true | | approve: true |\\n | } | | } |\\n +---------------------+ +---------------------+\\n |\\n v\\n status -> \\\"shutdown\\\", thread stops\\n\\n Plan approval FSM: pending -> approved | rejected\\n\\n Teammate Lead\\n +---------------------+ +---------------------+\\n | plan_approval | | |\\n | submit: {plan:\\\"...\\\"}| -------> | reviews plan text |\\n +---------------------+ | approve/reject? |\\n +---------------------+\\n |\\n +---------------------+ +-------v-------------+\\n | plan_approval_resp | <------- | plan_approval |\\n | {approve: true} | | review: {req_id, |\\n +---------------------+ | approve: true} |\\n +---------------------+\\n\\n Trackers: {request_id: {\\\"target|from\\\": name, \\\"status\\\": \\\"pending|...\\\"}}\\n\\nKey insight: \\\"Same request_id correlation pattern, two domains.\\\"\\n\\\"\\\"\\\"\\n\\nimport json\\nimport os\\nimport subprocess\\nimport threading\\nimport time\\nimport uuid\\nfrom pathlib import Path\\n\\nfrom anthropic import Anthropic\\nfrom dotenv import load_dotenv\\n\\nload_dotenv(override=True)\\nif os.getenv(\\\"ANTHROPIC_BASE_URL\\\"):\\n os.environ.pop(\\\"ANTHROPIC_AUTH_TOKEN\\\", None)\\n\\nWORKDIR = Path.cwd()\\nclient = Anthropic(base_url=os.getenv(\\\"ANTHROPIC_BASE_URL\\\"))\\nMODEL = os.environ[\\\"MODEL_ID\\\"]\\nTEAM_DIR = WORKDIR / \\\".team\\\"\\nINBOX_DIR = TEAM_DIR / \\\"inbox\\\"\\n\\nSYSTEM = f\\\"You are a team lead at {WORKDIR}. Manage teammates with shutdown and plan approval protocols.\\\"\\n\\nVALID_MSG_TYPES = {\\n \\\"message\\\",\\n \\\"broadcast\\\",\\n \\\"shutdown_request\\\",\\n \\\"shutdown_response\\\",\\n \\\"plan_approval_response\\\",\\n}\\n\\n# -- Request trackers: correlate by request_id --\\nshutdown_requests = {}\\nplan_requests = {}\\n_tracker_lock = threading.Lock()\\n\\n\\n# -- MessageBus: JSONL inbox per teammate --\\nclass MessageBus:\\n def __init__(self, inbox_dir: Path):\\n self.dir = inbox_dir\\n self.dir.mkdir(parents=True, exist_ok=True)\\n\\n def send(self, sender: str, to: str, content: str,\\n msg_type: str = \\\"message\\\", extra: dict = None) -> str:\\n if msg_type not in VALID_MSG_TYPES:\\n return f\\\"Error: Invalid type '{msg_type}'. Valid: {VALID_MSG_TYPES}\\\"\\n msg = {\\n \\\"type\\\": msg_type,\\n \\\"from\\\": sender,\\n \\\"content\\\": content,\\n \\\"timestamp\\\": time.time(),\\n }\\n if extra:\\n msg.update(extra)\\n inbox_path = self.dir / f\\\"{to}.jsonl\\\"\\n with open(inbox_path, \\\"a\\\") as f:\\n f.write(json.dumps(msg) + \\\"\\\\n\\\")\\n return f\\\"Sent {msg_type} to {to}\\\"\\n\\n def read_inbox(self, name: str) -> list:\\n inbox_path = self.dir / f\\\"{name}.jsonl\\\"\\n if not inbox_path.exists():\\n return []\\n messages = []\\n for line in inbox_path.read_text().strip().splitlines():\\n if line:\\n messages.append(json.loads(line))\\n inbox_path.write_text(\\\"\\\")\\n return messages\\n\\n def broadcast(self, sender: str, content: str, teammates: list) -> str:\\n count = 0\\n for name in teammates:\\n if name != sender:\\n self.send(sender, name, content, \\\"broadcast\\\")\\n count += 1\\n return f\\\"Broadcast to {count} teammates\\\"\\n\\n\\nBUS = MessageBus(INBOX_DIR)\\n\\n\\n# -- TeammateManager with shutdown + plan approval --\\nclass TeammateManager:\\n def __init__(self, team_dir: Path):\\n self.dir = team_dir\\n self.dir.mkdir(exist_ok=True)\\n self.config_path = self.dir / \\\"config.json\\\"\\n self.config = self._load_config()\\n self.threads = {}\\n\\n def _load_config(self) -> dict:\\n if self.config_path.exists():\\n return json.loads(self.config_path.read_text())\\n return {\\\"team_name\\\": \\\"default\\\", \\\"members\\\": []}\\n\\n def _save_config(self):\\n self.config_path.write_text(json.dumps(self.config, indent=2))\\n\\n def _find_member(self, name: str) -> dict:\\n for m in self.config[\\\"members\\\"]:\\n if m[\\\"name\\\"] == name:\\n return m\\n return None\\n\\n def spawn(self, name: str, role: str, prompt: str) -> str:\\n member = self._find_member(name)\\n if member:\\n if member[\\\"status\\\"] not in (\\\"idle\\\", \\\"shutdown\\\"):\\n return f\\\"Error: '{name}' is currently {member['status']}\\\"\\n member[\\\"status\\\"] = \\\"working\\\"\\n member[\\\"role\\\"] = role\\n else:\\n member = {\\\"name\\\": name, \\\"role\\\": role, \\\"status\\\": \\\"working\\\"}\\n self.config[\\\"members\\\"].append(member)\\n self._save_config()\\n thread = threading.Thread(\\n target=self._teammate_loop,\\n args=(name, role, prompt),\\n daemon=True,\\n )\\n self.threads[name] = thread\\n thread.start()\\n return f\\\"Spawned '{name}' (role: {role})\\\"\\n\\n def _teammate_loop(self, name: str, role: str, prompt: str):\\n sys_prompt = (\\n f\\\"You are '{name}', role: {role}, at {WORKDIR}. \\\"\\n f\\\"Submit plans via plan_approval before major work. \\\"\\n f\\\"Respond to shutdown_request with shutdown_response.\\\"\\n )\\n messages = [{\\\"role\\\": \\\"user\\\", \\\"content\\\": prompt}]\\n tools = self._teammate_tools()\\n should_exit = False\\n for _ in range(50):\\n inbox = BUS.read_inbox(name)\\n for msg in inbox:\\n messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": json.dumps(msg)})\\n if should_exit:\\n break\\n try:\\n response = client.messages.create(\\n model=MODEL,\\n system=sys_prompt,\\n messages=messages,\\n tools=tools,\\n max_tokens=8000,\\n )\\n except Exception:\\n break\\n messages.append({\\\"role\\\": \\\"assistant\\\", \\\"content\\\": response.content})\\n if response.stop_reason != \\\"tool_use\\\":\\n break\\n results = []\\n for block in response.content:\\n if block.type == \\\"tool_use\\\":\\n output = self._exec(name, block.name, block.input)\\n print(f\\\" [{name}] {block.name}: {str(output)[:120]}\\\")\\n results.append({\\n \\\"type\\\": \\\"tool_result\\\",\\n \\\"tool_use_id\\\": block.id,\\n \\\"content\\\": str(output),\\n })\\n if block.name == \\\"shutdown_response\\\" and block.input.get(\\\"approve\\\"):\\n should_exit = True\\n messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": results})\\n member = self._find_member(name)\\n if member:\\n member[\\\"status\\\"] = \\\"shutdown\\\" if should_exit else \\\"idle\\\"\\n self._save_config()\\n\\n def _exec(self, sender: str, tool_name: str, args: dict) -> str:\\n # these base tools are unchanged from s02\\n if tool_name == \\\"bash\\\":\\n return _run_bash(args[\\\"command\\\"])\\n if tool_name == \\\"read_file\\\":\\n return _run_read(args[\\\"path\\\"])\\n if tool_name == \\\"write_file\\\":\\n return _run_write(args[\\\"path\\\"], args[\\\"content\\\"])\\n if tool_name == \\\"edit_file\\\":\\n return _run_edit(args[\\\"path\\\"], args[\\\"old_text\\\"], args[\\\"new_text\\\"])\\n if tool_name == \\\"send_message\\\":\\n return BUS.send(sender, args[\\\"to\\\"], args[\\\"content\\\"], args.get(\\\"msg_type\\\", \\\"message\\\"))\\n if tool_name == \\\"read_inbox\\\":\\n return json.dumps(BUS.read_inbox(sender), indent=2)\\n if tool_name == \\\"shutdown_response\\\":\\n req_id = args[\\\"request_id\\\"]\\n approve = args[\\\"approve\\\"]\\n with _tracker_lock:\\n if req_id in shutdown_requests:\\n shutdown_requests[req_id][\\\"status\\\"] = \\\"approved\\\" if approve else \\\"rejected\\\"\\n BUS.send(\\n sender, \\\"lead\\\", args.get(\\\"reason\\\", \\\"\\\"),\\n \\\"shutdown_response\\\", {\\\"request_id\\\": req_id, \\\"approve\\\": approve},\\n )\\n return f\\\"Shutdown {'approved' if approve else 'rejected'}\\\"\\n if tool_name == \\\"plan_approval\\\":\\n plan_text = args.get(\\\"plan\\\", \\\"\\\")\\n req_id = str(uuid.uuid4())[:8]\\n with _tracker_lock:\\n plan_requests[req_id] = {\\\"from\\\": sender, \\\"plan\\\": plan_text, \\\"status\\\": \\\"pending\\\"}\\n BUS.send(\\n sender, \\\"lead\\\", plan_text, \\\"plan_approval_response\\\",\\n {\\\"request_id\\\": req_id, \\\"plan\\\": plan_text},\\n )\\n return f\\\"Plan submitted (request_id={req_id}). Waiting for lead approval.\\\"\\n return f\\\"Unknown tool: {tool_name}\\\"\\n\\n def _teammate_tools(self) -> list:\\n # these base tools are unchanged from s02\\n return [\\n {\\\"name\\\": \\\"bash\\\", \\\"description\\\": \\\"Run a shell command.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"command\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"command\\\"]}},\\n {\\\"name\\\": \\\"read_file\\\", \\\"description\\\": \\\"Read file contents.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\"]}},\\n {\\\"name\\\": \\\"write_file\\\", \\\"description\\\": \\\"Write content to file.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"content\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\", \\\"content\\\"]}},\\n {\\\"name\\\": \\\"edit_file\\\", \\\"description\\\": \\\"Replace exact text in file.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"old_text\\\": {\\\"type\\\": \\\"string\\\"}, \\\"new_text\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\", \\\"old_text\\\", \\\"new_text\\\"]}},\\n {\\\"name\\\": \\\"send_message\\\", \\\"description\\\": \\\"Send message to a teammate.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"to\\\": {\\\"type\\\": \\\"string\\\"}, \\\"content\\\": {\\\"type\\\": \\\"string\\\"}, \\\"msg_type\\\": {\\\"type\\\": \\\"string\\\", \\\"enum\\\": list(VALID_MSG_TYPES)}}, \\\"required\\\": [\\\"to\\\", \\\"content\\\"]}},\\n {\\\"name\\\": \\\"read_inbox\\\", \\\"description\\\": \\\"Read and drain your inbox.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {}}},\\n {\\\"name\\\": \\\"shutdown_response\\\", \\\"description\\\": \\\"Respond to a shutdown request. Approve to shut down, reject to keep working.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"request_id\\\": {\\\"type\\\": \\\"string\\\"}, \\\"approve\\\": {\\\"type\\\": \\\"boolean\\\"}, \\\"reason\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"request_id\\\", \\\"approve\\\"]}},\\n {\\\"name\\\": \\\"plan_approval\\\", \\\"description\\\": \\\"Submit a plan for lead approval. Provide plan text.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"plan\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"plan\\\"]}},\\n ]\\n\\n def list_all(self) -> str:\\n if not self.config[\\\"members\\\"]:\\n return \\\"No teammates.\\\"\\n lines = [f\\\"Team: {self.config['team_name']}\\\"]\\n for m in self.config[\\\"members\\\"]:\\n lines.append(f\\\" {m['name']} ({m['role']}): {m['status']}\\\")\\n return \\\"\\\\n\\\".join(lines)\\n\\n def member_names(self) -> list:\\n return [m[\\\"name\\\"] for m in self.config[\\\"members\\\"]]\\n\\n\\nTEAM = TeammateManager(TEAM_DIR)\\n\\n\\n# -- Base tool implementations (these base tools are unchanged from s02) --\\ndef _safe_path(p: str) -> Path:\\n path = (WORKDIR / p).resolve()\\n if not path.is_relative_to(WORKDIR):\\n raise ValueError(f\\\"Path escapes workspace: {p}\\\")\\n return path\\n\\n\\ndef _run_bash(command: str) -> str:\\n dangerous = [\\\"rm -rf /\\\", \\\"sudo\\\", \\\"shutdown\\\", \\\"reboot\\\"]\\n if any(d in command for d in dangerous):\\n return \\\"Error: Dangerous command blocked\\\"\\n try:\\n r = subprocess.run(\\n command, shell=True, cwd=WORKDIR,\\n capture_output=True, text=True, timeout=120,\\n )\\n out = (r.stdout + r.stderr).strip()\\n return out[:50000] if out else \\\"(no output)\\\"\\n except subprocess.TimeoutExpired:\\n return \\\"Error: Timeout (120s)\\\"\\n\\n\\ndef _run_read(path: str, limit: int = None) -> str:\\n try:\\n lines = _safe_path(path).read_text().splitlines()\\n if limit and limit < len(lines):\\n lines = lines[:limit] + [f\\\"... ({len(lines) - limit} more)\\\"]\\n return \\\"\\\\n\\\".join(lines)[:50000]\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\n\\ndef _run_write(path: str, content: str) -> str:\\n try:\\n fp = _safe_path(path)\\n fp.parent.mkdir(parents=True, exist_ok=True)\\n fp.write_text(content)\\n return f\\\"Wrote {len(content)} bytes\\\"\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\n\\ndef _run_edit(path: str, old_text: str, new_text: str) -> str:\\n try:\\n fp = _safe_path(path)\\n c = fp.read_text()\\n if old_text not in c:\\n return f\\\"Error: Text not found in {path}\\\"\\n fp.write_text(c.replace(old_text, new_text, 1))\\n return f\\\"Edited {path}\\\"\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\n\\n# -- Lead-specific protocol handlers --\\ndef handle_shutdown_request(teammate: str) -> str:\\n req_id = str(uuid.uuid4())[:8]\\n with _tracker_lock:\\n shutdown_requests[req_id] = {\\\"target\\\": teammate, \\\"status\\\": \\\"pending\\\"}\\n BUS.send(\\n \\\"lead\\\", teammate, \\\"Please shut down gracefully.\\\",\\n \\\"shutdown_request\\\", {\\\"request_id\\\": req_id},\\n )\\n return f\\\"Shutdown request {req_id} sent to '{teammate}' (status: pending)\\\"\\n\\n\\ndef handle_plan_review(request_id: str, approve: bool, feedback: str = \\\"\\\") -> str:\\n with _tracker_lock:\\n req = plan_requests.get(request_id)\\n if not req:\\n return f\\\"Error: Unknown plan request_id '{request_id}'\\\"\\n with _tracker_lock:\\n req[\\\"status\\\"] = \\\"approved\\\" if approve else \\\"rejected\\\"\\n BUS.send(\\n \\\"lead\\\", req[\\\"from\\\"], feedback, \\\"plan_approval_response\\\",\\n {\\\"request_id\\\": request_id, \\\"approve\\\": approve, \\\"feedback\\\": feedback},\\n )\\n return f\\\"Plan {req['status']} for '{req['from']}'\\\"\\n\\n\\ndef _check_shutdown_status(request_id: str) -> str:\\n with _tracker_lock:\\n return json.dumps(shutdown_requests.get(request_id, {\\\"error\\\": \\\"not found\\\"}))\\n\\n\\n# -- Lead tool dispatch (12 tools) --\\nTOOL_HANDLERS = {\\n \\\"bash\\\": lambda **kw: _run_bash(kw[\\\"command\\\"]),\\n \\\"read_file\\\": lambda **kw: _run_read(kw[\\\"path\\\"], kw.get(\\\"limit\\\")),\\n \\\"write_file\\\": lambda **kw: _run_write(kw[\\\"path\\\"], kw[\\\"content\\\"]),\\n \\\"edit_file\\\": lambda **kw: _run_edit(kw[\\\"path\\\"], kw[\\\"old_text\\\"], kw[\\\"new_text\\\"]),\\n \\\"spawn_teammate\\\": lambda **kw: TEAM.spawn(kw[\\\"name\\\"], kw[\\\"role\\\"], kw[\\\"prompt\\\"]),\\n \\\"list_teammates\\\": lambda **kw: TEAM.list_all(),\\n \\\"send_message\\\": lambda **kw: BUS.send(\\\"lead\\\", kw[\\\"to\\\"], kw[\\\"content\\\"], kw.get(\\\"msg_type\\\", \\\"message\\\")),\\n \\\"read_inbox\\\": lambda **kw: json.dumps(BUS.read_inbox(\\\"lead\\\"), indent=2),\\n \\\"broadcast\\\": lambda **kw: BUS.broadcast(\\\"lead\\\", kw[\\\"content\\\"], TEAM.member_names()),\\n \\\"shutdown_request\\\": lambda **kw: handle_shutdown_request(kw[\\\"teammate\\\"]),\\n \\\"shutdown_response\\\": lambda **kw: _check_shutdown_status(kw.get(\\\"request_id\\\", \\\"\\\")),\\n \\\"plan_approval\\\": lambda **kw: handle_plan_review(kw[\\\"request_id\\\"], kw[\\\"approve\\\"], kw.get(\\\"feedback\\\", \\\"\\\")),\\n}\\n\\n# these base tools are unchanged from s02\\nTOOLS = [\\n {\\\"name\\\": \\\"bash\\\", \\\"description\\\": \\\"Run a shell command.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"command\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"command\\\"]}},\\n {\\\"name\\\": \\\"read_file\\\", \\\"description\\\": \\\"Read file contents.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"limit\\\": {\\\"type\\\": \\\"integer\\\"}}, \\\"required\\\": [\\\"path\\\"]}},\\n {\\\"name\\\": \\\"write_file\\\", \\\"description\\\": \\\"Write content to file.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"content\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\", \\\"content\\\"]}},\\n {\\\"name\\\": \\\"edit_file\\\", \\\"description\\\": \\\"Replace exact text in file.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"old_text\\\": {\\\"type\\\": \\\"string\\\"}, \\\"new_text\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\", \\\"old_text\\\", \\\"new_text\\\"]}},\\n {\\\"name\\\": \\\"spawn_teammate\\\", \\\"description\\\": \\\"Spawn a persistent teammate.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"name\\\": {\\\"type\\\": \\\"string\\\"}, \\\"role\\\": {\\\"type\\\": \\\"string\\\"}, \\\"prompt\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"name\\\", \\\"role\\\", \\\"prompt\\\"]}},\\n {\\\"name\\\": \\\"list_teammates\\\", \\\"description\\\": \\\"List all teammates.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {}}},\\n {\\\"name\\\": \\\"send_message\\\", \\\"description\\\": \\\"Send a message to a teammate.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"to\\\": {\\\"type\\\": \\\"string\\\"}, \\\"content\\\": {\\\"type\\\": \\\"string\\\"}, \\\"msg_type\\\": {\\\"type\\\": \\\"string\\\", \\\"enum\\\": list(VALID_MSG_TYPES)}}, \\\"required\\\": [\\\"to\\\", \\\"content\\\"]}},\\n {\\\"name\\\": \\\"read_inbox\\\", \\\"description\\\": \\\"Read and drain the lead's inbox.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {}}},\\n {\\\"name\\\": \\\"broadcast\\\", \\\"description\\\": \\\"Send a message to all teammates.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"content\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"content\\\"]}},\\n {\\\"name\\\": \\\"shutdown_request\\\", \\\"description\\\": \\\"Request a teammate to shut down gracefully. Returns a request_id for tracking.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"teammate\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"teammate\\\"]}},\\n {\\\"name\\\": \\\"shutdown_response\\\", \\\"description\\\": \\\"Check the status of a shutdown request by request_id.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"request_id\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"request_id\\\"]}},\\n {\\\"name\\\": \\\"plan_approval\\\", \\\"description\\\": \\\"Approve or reject a teammate's plan. Provide request_id + approve + optional feedback.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"request_id\\\": {\\\"type\\\": \\\"string\\\"}, \\\"approve\\\": {\\\"type\\\": \\\"boolean\\\"}, \\\"feedback\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"request_id\\\", \\\"approve\\\"]}},\\n]\\n\\n\\ndef agent_loop(messages: list):\\n while True:\\n inbox = BUS.read_inbox(\\\"lead\\\")\\n if inbox:\\n messages.append({\\n \\\"role\\\": \\\"user\\\",\\n \\\"content\\\": f\\\"{json.dumps(inbox, indent=2)} \\\",\\n })\\n messages.append({\\n \\\"role\\\": \\\"assistant\\\",\\n \\\"content\\\": \\\"Noted inbox messages.\\\",\\n })\\n response = client.messages.create(\\n model=MODEL,\\n system=SYSTEM,\\n messages=messages,\\n tools=TOOLS,\\n max_tokens=8000,\\n )\\n messages.append({\\\"role\\\": \\\"assistant\\\", \\\"content\\\": response.content})\\n if response.stop_reason != \\\"tool_use\\\":\\n return\\n results = []\\n for block in response.content:\\n if block.type == \\\"tool_use\\\":\\n handler = TOOL_HANDLERS.get(block.name)\\n try:\\n output = handler(**block.input) if handler else f\\\"Unknown tool: {block.name}\\\"\\n except Exception as e:\\n output = f\\\"Error: {e}\\\"\\n print(f\\\"> {block.name}: {str(output)[:200]}\\\")\\n results.append({\\n \\\"type\\\": \\\"tool_result\\\",\\n \\\"tool_use_id\\\": block.id,\\n \\\"content\\\": str(output),\\n })\\n messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": results})\\n\\n\\nif __name__ == \\\"__main__\\\":\\n history = []\\n while True:\\n try:\\n query = input(\\\"\\\\033[36ms10 >> \\\\033[0m\\\")\\n except (EOFError, KeyboardInterrupt):\\n break\\n if query.strip().lower() in (\\\"q\\\", \\\"exit\\\", \\\"\\\"):\\n break\\n if query.strip() == \\\"/team\\\":\\n print(TEAM.list_all())\\n continue\\n if query.strip() == \\\"/inbox\\\":\\n print(json.dumps(BUS.read_inbox(\\\"lead\\\"), indent=2))\\n continue\\n history.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": query})\\n agent_loop(history)\\n print()\\n\"\n },\n {\n \"id\": \"s11\",\n \"filename\": \"s11_autonomous_agents.py\",\n \"title\": \"Autonomous Agents\",\n \"subtitle\": \"Scan Board, Claim Tasks\",\n \"loc\": 494,\n \"tools\": [\n \"bash\",\n \"read_file\",\n \"write_file\",\n \"edit_file\",\n \"send_message\",\n \"read_inbox\",\n \"shutdown_response\",\n \"plan_approval\",\n \"idle\",\n \"claim_task\",\n \"spawn_teammate\",\n \"list_teammates\",\n \"broadcast\",\n \"shutdown_request\"\n ],\n \"newTools\": [\n \"idle\",\n \"claim_task\"\n ],\n \"coreAddition\": \"Task board polling + timeout-based self-governance\",\n \"keyInsight\": \"Teammates scan the board and claim tasks themselves; no need for the lead to assign each one\",\n \"classes\": [\n {\n \"name\": \"MessageBus\",\n \"startLine\": 80,\n \"endLine\": 121\n },\n {\n \"name\": \"TeammateManager\",\n \"startLine\": 159,\n \"endLine\": 368\n }\n ],\n \"functions\": [\n {\n \"name\": \"scan_unclaimed_tasks\",\n \"signature\": \"def scan_unclaimed_tasks()\",\n \"startLine\": 126\n },\n {\n \"name\": \"claim_task\",\n \"signature\": \"def claim_task(task_id: int, owner: str)\",\n \"startLine\": 138\n },\n {\n \"name\": \"make_identity_block\",\n \"signature\": \"def make_identity_block(name: str, role: str, team_name: str)\",\n \"startLine\": 151\n },\n {\n \"name\": \"_safe_path\",\n \"signature\": \"def _safe_path(p: str)\",\n \"startLine\": 373\n },\n {\n \"name\": \"_run_bash\",\n \"signature\": \"def _run_bash(command: str)\",\n \"startLine\": 380\n },\n {\n \"name\": \"_run_read\",\n \"signature\": \"def _run_read(path: str, limit: int = None)\",\n \"startLine\": 395\n },\n {\n \"name\": \"_run_write\",\n \"signature\": \"def _run_write(path: str, content: str)\",\n \"startLine\": 405\n },\n {\n \"name\": \"_run_edit\",\n \"signature\": \"def _run_edit(path: str, old_text: str, new_text: str)\",\n \"startLine\": 415\n },\n {\n \"name\": \"handle_shutdown_request\",\n \"signature\": \"def handle_shutdown_request(teammate: str)\",\n \"startLine\": 428\n },\n {\n \"name\": \"handle_plan_review\",\n \"signature\": \"def handle_plan_review(request_id: str, approve: bool, feedback: str = \\\"\\\")\",\n \"startLine\": 439\n },\n {\n \"name\": \"_check_shutdown_status\",\n \"signature\": \"def _check_shutdown_status(request_id: str)\",\n \"startLine\": 453\n },\n {\n \"name\": \"agent_loop\",\n \"signature\": \"def agent_loop(messages: list)\",\n \"startLine\": 509\n }\n ],\n \"layer\": \"collaboration\",\n \"source\": \"#!/usr/bin/env python3\\n\\\"\\\"\\\"\\ns11_autonomous_agents.py - Autonomous Agents\\n\\nIdle cycle with task board polling, auto-claiming unclaimed tasks, and\\nidentity re-injection after context compression. Builds on s10's protocols.\\n\\n Teammate lifecycle:\\n +-------+\\n | spawn |\\n +---+---+\\n |\\n v\\n +-------+ tool_use +-------+\\n | WORK | <----------- | LLM |\\n +---+---+ +-------+\\n |\\n | stop_reason != tool_use\\n v\\n +--------+\\n | IDLE | poll every 5s for up to 60s\\n +---+----+\\n |\\n +---> check inbox -> message? -> resume WORK\\n |\\n +---> scan .tasks/ -> unclaimed? -> claim -> resume WORK\\n |\\n +---> timeout (60s) -> shutdown\\n\\n Identity re-injection after compression:\\n messages = [identity_block, ...remaining...]\\n \\\"You are 'coder', role: backend, team: my-team\\\"\\n\\nKey insight: \\\"The agent finds work itself.\\\"\\n\\\"\\\"\\\"\\n\\nimport json\\nimport os\\nimport subprocess\\nimport threading\\nimport time\\nimport uuid\\nfrom pathlib import Path\\n\\nfrom anthropic import Anthropic\\nfrom dotenv import load_dotenv\\n\\nload_dotenv(override=True)\\nif os.getenv(\\\"ANTHROPIC_BASE_URL\\\"):\\n os.environ.pop(\\\"ANTHROPIC_AUTH_TOKEN\\\", None)\\n\\nWORKDIR = Path.cwd()\\nclient = Anthropic(base_url=os.getenv(\\\"ANTHROPIC_BASE_URL\\\"))\\nMODEL = os.environ[\\\"MODEL_ID\\\"]\\nTEAM_DIR = WORKDIR / \\\".team\\\"\\nINBOX_DIR = TEAM_DIR / \\\"inbox\\\"\\nTASKS_DIR = WORKDIR / \\\".tasks\\\"\\n\\nPOLL_INTERVAL = 5\\nIDLE_TIMEOUT = 60\\n\\nSYSTEM = f\\\"You are a team lead at {WORKDIR}. Teammates are autonomous -- they find work themselves.\\\"\\n\\nVALID_MSG_TYPES = {\\n \\\"message\\\",\\n \\\"broadcast\\\",\\n \\\"shutdown_request\\\",\\n \\\"shutdown_response\\\",\\n \\\"plan_approval_response\\\",\\n}\\n\\n# -- Request trackers --\\nshutdown_requests = {}\\nplan_requests = {}\\n_tracker_lock = threading.Lock()\\n_claim_lock = threading.Lock()\\n\\n\\n# -- MessageBus: JSONL inbox per teammate --\\nclass MessageBus:\\n def __init__(self, inbox_dir: Path):\\n self.dir = inbox_dir\\n self.dir.mkdir(parents=True, exist_ok=True)\\n\\n def send(self, sender: str, to: str, content: str,\\n msg_type: str = \\\"message\\\", extra: dict = None) -> str:\\n if msg_type not in VALID_MSG_TYPES:\\n return f\\\"Error: Invalid type '{msg_type}'. Valid: {VALID_MSG_TYPES}\\\"\\n msg = {\\n \\\"type\\\": msg_type,\\n \\\"from\\\": sender,\\n \\\"content\\\": content,\\n \\\"timestamp\\\": time.time(),\\n }\\n if extra:\\n msg.update(extra)\\n inbox_path = self.dir / f\\\"{to}.jsonl\\\"\\n with open(inbox_path, \\\"a\\\") as f:\\n f.write(json.dumps(msg) + \\\"\\\\n\\\")\\n return f\\\"Sent {msg_type} to {to}\\\"\\n\\n def read_inbox(self, name: str) -> list:\\n inbox_path = self.dir / f\\\"{name}.jsonl\\\"\\n if not inbox_path.exists():\\n return []\\n messages = []\\n for line in inbox_path.read_text().strip().splitlines():\\n if line:\\n messages.append(json.loads(line))\\n inbox_path.write_text(\\\"\\\")\\n return messages\\n\\n def broadcast(self, sender: str, content: str, teammates: list) -> str:\\n count = 0\\n for name in teammates:\\n if name != sender:\\n self.send(sender, name, content, \\\"broadcast\\\")\\n count += 1\\n return f\\\"Broadcast to {count} teammates\\\"\\n\\n\\nBUS = MessageBus(INBOX_DIR)\\n\\n\\n# -- Task board scanning --\\ndef scan_unclaimed_tasks() -> list:\\n TASKS_DIR.mkdir(exist_ok=True)\\n unclaimed = []\\n for f in sorted(TASKS_DIR.glob(\\\"task_*.json\\\")):\\n task = json.loads(f.read_text())\\n if (task.get(\\\"status\\\") == \\\"pending\\\"\\n and not task.get(\\\"owner\\\")\\n and not task.get(\\\"blockedBy\\\")):\\n unclaimed.append(task)\\n return unclaimed\\n\\n\\ndef claim_task(task_id: int, owner: str) -> str:\\n with _claim_lock:\\n path = TASKS_DIR / f\\\"task_{task_id}.json\\\"\\n if not path.exists():\\n return f\\\"Error: Task {task_id} not found\\\"\\n task = json.loads(path.read_text())\\n task[\\\"owner\\\"] = owner\\n task[\\\"status\\\"] = \\\"in_progress\\\"\\n path.write_text(json.dumps(task, indent=2))\\n return f\\\"Claimed task #{task_id} for {owner}\\\"\\n\\n\\n# -- Identity re-injection after compression --\\ndef make_identity_block(name: str, role: str, team_name: str) -> dict:\\n return {\\n \\\"role\\\": \\\"user\\\",\\n \\\"content\\\": f\\\"You are '{name}', role: {role}, team: {team_name}. Continue your work. \\\",\\n }\\n\\n\\n# -- Autonomous TeammateManager --\\nclass TeammateManager:\\n def __init__(self, team_dir: Path):\\n self.dir = team_dir\\n self.dir.mkdir(exist_ok=True)\\n self.config_path = self.dir / \\\"config.json\\\"\\n self.config = self._load_config()\\n self.threads = {}\\n\\n def _load_config(self) -> dict:\\n if self.config_path.exists():\\n return json.loads(self.config_path.read_text())\\n return {\\\"team_name\\\": \\\"default\\\", \\\"members\\\": []}\\n\\n def _save_config(self):\\n self.config_path.write_text(json.dumps(self.config, indent=2))\\n\\n def _find_member(self, name: str) -> dict:\\n for m in self.config[\\\"members\\\"]:\\n if m[\\\"name\\\"] == name:\\n return m\\n return None\\n\\n def _set_status(self, name: str, status: str):\\n member = self._find_member(name)\\n if member:\\n member[\\\"status\\\"] = status\\n self._save_config()\\n\\n def spawn(self, name: str, role: str, prompt: str) -> str:\\n member = self._find_member(name)\\n if member:\\n if member[\\\"status\\\"] not in (\\\"idle\\\", \\\"shutdown\\\"):\\n return f\\\"Error: '{name}' is currently {member['status']}\\\"\\n member[\\\"status\\\"] = \\\"working\\\"\\n member[\\\"role\\\"] = role\\n else:\\n member = {\\\"name\\\": name, \\\"role\\\": role, \\\"status\\\": \\\"working\\\"}\\n self.config[\\\"members\\\"].append(member)\\n self._save_config()\\n thread = threading.Thread(\\n target=self._loop,\\n args=(name, role, prompt),\\n daemon=True,\\n )\\n self.threads[name] = thread\\n thread.start()\\n return f\\\"Spawned '{name}' (role: {role})\\\"\\n\\n def _loop(self, name: str, role: str, prompt: str):\\n team_name = self.config[\\\"team_name\\\"]\\n sys_prompt = (\\n f\\\"You are '{name}', role: {role}, team: {team_name}, at {WORKDIR}. \\\"\\n f\\\"Use idle tool when you have no more work. You will auto-claim new tasks.\\\"\\n )\\n messages = [{\\\"role\\\": \\\"user\\\", \\\"content\\\": prompt}]\\n tools = self._teammate_tools()\\n\\n while True:\\n # -- WORK PHASE: standard agent loop --\\n for _ in range(50):\\n inbox = BUS.read_inbox(name)\\n for msg in inbox:\\n if msg.get(\\\"type\\\") == \\\"shutdown_request\\\":\\n self._set_status(name, \\\"shutdown\\\")\\n return\\n messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": json.dumps(msg)})\\n try:\\n response = client.messages.create(\\n model=MODEL,\\n system=sys_prompt,\\n messages=messages,\\n tools=tools,\\n max_tokens=8000,\\n )\\n except Exception:\\n self._set_status(name, \\\"idle\\\")\\n return\\n messages.append({\\\"role\\\": \\\"assistant\\\", \\\"content\\\": response.content})\\n if response.stop_reason != \\\"tool_use\\\":\\n break\\n results = []\\n idle_requested = False\\n for block in response.content:\\n if block.type == \\\"tool_use\\\":\\n if block.name == \\\"idle\\\":\\n idle_requested = True\\n output = \\\"Entering idle phase. Will poll for new tasks.\\\"\\n else:\\n output = self._exec(name, block.name, block.input)\\n print(f\\\" [{name}] {block.name}: {str(output)[:120]}\\\")\\n results.append({\\n \\\"type\\\": \\\"tool_result\\\",\\n \\\"tool_use_id\\\": block.id,\\n \\\"content\\\": str(output),\\n })\\n messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": results})\\n if idle_requested:\\n break\\n\\n # -- IDLE PHASE: poll for inbox messages and unclaimed tasks --\\n self._set_status(name, \\\"idle\\\")\\n resume = False\\n polls = IDLE_TIMEOUT // max(POLL_INTERVAL, 1)\\n for _ in range(polls):\\n time.sleep(POLL_INTERVAL)\\n inbox = BUS.read_inbox(name)\\n if inbox:\\n for msg in inbox:\\n if msg.get(\\\"type\\\") == \\\"shutdown_request\\\":\\n self._set_status(name, \\\"shutdown\\\")\\n return\\n messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": json.dumps(msg)})\\n resume = True\\n break\\n unclaimed = scan_unclaimed_tasks()\\n if unclaimed:\\n task = unclaimed[0]\\n claim_task(task[\\\"id\\\"], name)\\n task_prompt = (\\n f\\\"Task #{task['id']}: {task['subject']}\\\\n\\\"\\n f\\\"{task.get('description', '')} \\\"\\n )\\n if len(messages) <= 3:\\n messages.insert(0, make_identity_block(name, role, team_name))\\n messages.insert(1, {\\\"role\\\": \\\"assistant\\\", \\\"content\\\": f\\\"I am {name}. Continuing.\\\"})\\n messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": task_prompt})\\n messages.append({\\\"role\\\": \\\"assistant\\\", \\\"content\\\": f\\\"Claimed task #{task['id']}. Working on it.\\\"})\\n resume = True\\n break\\n\\n if not resume:\\n self._set_status(name, \\\"shutdown\\\")\\n return\\n self._set_status(name, \\\"working\\\")\\n\\n def _exec(self, sender: str, tool_name: str, args: dict) -> str:\\n # these base tools are unchanged from s02\\n if tool_name == \\\"bash\\\":\\n return _run_bash(args[\\\"command\\\"])\\n if tool_name == \\\"read_file\\\":\\n return _run_read(args[\\\"path\\\"])\\n if tool_name == \\\"write_file\\\":\\n return _run_write(args[\\\"path\\\"], args[\\\"content\\\"])\\n if tool_name == \\\"edit_file\\\":\\n return _run_edit(args[\\\"path\\\"], args[\\\"old_text\\\"], args[\\\"new_text\\\"])\\n if tool_name == \\\"send_message\\\":\\n return BUS.send(sender, args[\\\"to\\\"], args[\\\"content\\\"], args.get(\\\"msg_type\\\", \\\"message\\\"))\\n if tool_name == \\\"read_inbox\\\":\\n return json.dumps(BUS.read_inbox(sender), indent=2)\\n if tool_name == \\\"shutdown_response\\\":\\n req_id = args[\\\"request_id\\\"]\\n with _tracker_lock:\\n if req_id in shutdown_requests:\\n shutdown_requests[req_id][\\\"status\\\"] = \\\"approved\\\" if args[\\\"approve\\\"] else \\\"rejected\\\"\\n BUS.send(\\n sender, \\\"lead\\\", args.get(\\\"reason\\\", \\\"\\\"),\\n \\\"shutdown_response\\\", {\\\"request_id\\\": req_id, \\\"approve\\\": args[\\\"approve\\\"]},\\n )\\n return f\\\"Shutdown {'approved' if args['approve'] else 'rejected'}\\\"\\n if tool_name == \\\"plan_approval\\\":\\n plan_text = args.get(\\\"plan\\\", \\\"\\\")\\n req_id = str(uuid.uuid4())[:8]\\n with _tracker_lock:\\n plan_requests[req_id] = {\\\"from\\\": sender, \\\"plan\\\": plan_text, \\\"status\\\": \\\"pending\\\"}\\n BUS.send(\\n sender, \\\"lead\\\", plan_text, \\\"plan_approval_response\\\",\\n {\\\"request_id\\\": req_id, \\\"plan\\\": plan_text},\\n )\\n return f\\\"Plan submitted (request_id={req_id}). Waiting for approval.\\\"\\n if tool_name == \\\"claim_task\\\":\\n return claim_task(args[\\\"task_id\\\"], sender)\\n return f\\\"Unknown tool: {tool_name}\\\"\\n\\n def _teammate_tools(self) -> list:\\n # these base tools are unchanged from s02\\n return [\\n {\\\"name\\\": \\\"bash\\\", \\\"description\\\": \\\"Run a shell command.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"command\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"command\\\"]}},\\n {\\\"name\\\": \\\"read_file\\\", \\\"description\\\": \\\"Read file contents.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\"]}},\\n {\\\"name\\\": \\\"write_file\\\", \\\"description\\\": \\\"Write content to file.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"content\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\", \\\"content\\\"]}},\\n {\\\"name\\\": \\\"edit_file\\\", \\\"description\\\": \\\"Replace exact text in file.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"old_text\\\": {\\\"type\\\": \\\"string\\\"}, \\\"new_text\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\", \\\"old_text\\\", \\\"new_text\\\"]}},\\n {\\\"name\\\": \\\"send_message\\\", \\\"description\\\": \\\"Send message to a teammate.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"to\\\": {\\\"type\\\": \\\"string\\\"}, \\\"content\\\": {\\\"type\\\": \\\"string\\\"}, \\\"msg_type\\\": {\\\"type\\\": \\\"string\\\", \\\"enum\\\": list(VALID_MSG_TYPES)}}, \\\"required\\\": [\\\"to\\\", \\\"content\\\"]}},\\n {\\\"name\\\": \\\"read_inbox\\\", \\\"description\\\": \\\"Read and drain your inbox.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {}}},\\n {\\\"name\\\": \\\"shutdown_response\\\", \\\"description\\\": \\\"Respond to a shutdown request.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"request_id\\\": {\\\"type\\\": \\\"string\\\"}, \\\"approve\\\": {\\\"type\\\": \\\"boolean\\\"}, \\\"reason\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"request_id\\\", \\\"approve\\\"]}},\\n {\\\"name\\\": \\\"plan_approval\\\", \\\"description\\\": \\\"Submit a plan for lead approval.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"plan\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"plan\\\"]}},\\n {\\\"name\\\": \\\"idle\\\", \\\"description\\\": \\\"Signal that you have no more work. Enters idle polling phase.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {}}},\\n {\\\"name\\\": \\\"claim_task\\\", \\\"description\\\": \\\"Claim a task from the task board by ID.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"task_id\\\": {\\\"type\\\": \\\"integer\\\"}}, \\\"required\\\": [\\\"task_id\\\"]}},\\n ]\\n\\n def list_all(self) -> str:\\n if not self.config[\\\"members\\\"]:\\n return \\\"No teammates.\\\"\\n lines = [f\\\"Team: {self.config['team_name']}\\\"]\\n for m in self.config[\\\"members\\\"]:\\n lines.append(f\\\" {m['name']} ({m['role']}): {m['status']}\\\")\\n return \\\"\\\\n\\\".join(lines)\\n\\n def member_names(self) -> list:\\n return [m[\\\"name\\\"] for m in self.config[\\\"members\\\"]]\\n\\n\\nTEAM = TeammateManager(TEAM_DIR)\\n\\n\\n# -- Base tool implementations (these base tools are unchanged from s02) --\\ndef _safe_path(p: str) -> Path:\\n path = (WORKDIR / p).resolve()\\n if not path.is_relative_to(WORKDIR):\\n raise ValueError(f\\\"Path escapes workspace: {p}\\\")\\n return path\\n\\n\\ndef _run_bash(command: str) -> str:\\n dangerous = [\\\"rm -rf /\\\", \\\"sudo\\\", \\\"shutdown\\\", \\\"reboot\\\"]\\n if any(d in command for d in dangerous):\\n return \\\"Error: Dangerous command blocked\\\"\\n try:\\n r = subprocess.run(\\n command, shell=True, cwd=WORKDIR,\\n capture_output=True, text=True, timeout=120,\\n )\\n out = (r.stdout + r.stderr).strip()\\n return out[:50000] if out else \\\"(no output)\\\"\\n except subprocess.TimeoutExpired:\\n return \\\"Error: Timeout (120s)\\\"\\n\\n\\ndef _run_read(path: str, limit: int = None) -> str:\\n try:\\n lines = _safe_path(path).read_text().splitlines()\\n if limit and limit < len(lines):\\n lines = lines[:limit] + [f\\\"... ({len(lines) - limit} more)\\\"]\\n return \\\"\\\\n\\\".join(lines)[:50000]\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\n\\ndef _run_write(path: str, content: str) -> str:\\n try:\\n fp = _safe_path(path)\\n fp.parent.mkdir(parents=True, exist_ok=True)\\n fp.write_text(content)\\n return f\\\"Wrote {len(content)} bytes\\\"\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\n\\ndef _run_edit(path: str, old_text: str, new_text: str) -> str:\\n try:\\n fp = _safe_path(path)\\n c = fp.read_text()\\n if old_text not in c:\\n return f\\\"Error: Text not found in {path}\\\"\\n fp.write_text(c.replace(old_text, new_text, 1))\\n return f\\\"Edited {path}\\\"\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\n\\n# -- Lead-specific protocol handlers --\\ndef handle_shutdown_request(teammate: str) -> str:\\n req_id = str(uuid.uuid4())[:8]\\n with _tracker_lock:\\n shutdown_requests[req_id] = {\\\"target\\\": teammate, \\\"status\\\": \\\"pending\\\"}\\n BUS.send(\\n \\\"lead\\\", teammate, \\\"Please shut down gracefully.\\\",\\n \\\"shutdown_request\\\", {\\\"request_id\\\": req_id},\\n )\\n return f\\\"Shutdown request {req_id} sent to '{teammate}'\\\"\\n\\n\\ndef handle_plan_review(request_id: str, approve: bool, feedback: str = \\\"\\\") -> str:\\n with _tracker_lock:\\n req = plan_requests.get(request_id)\\n if not req:\\n return f\\\"Error: Unknown plan request_id '{request_id}'\\\"\\n with _tracker_lock:\\n req[\\\"status\\\"] = \\\"approved\\\" if approve else \\\"rejected\\\"\\n BUS.send(\\n \\\"lead\\\", req[\\\"from\\\"], feedback, \\\"plan_approval_response\\\",\\n {\\\"request_id\\\": request_id, \\\"approve\\\": approve, \\\"feedback\\\": feedback},\\n )\\n return f\\\"Plan {req['status']} for '{req['from']}'\\\"\\n\\n\\ndef _check_shutdown_status(request_id: str) -> str:\\n with _tracker_lock:\\n return json.dumps(shutdown_requests.get(request_id, {\\\"error\\\": \\\"not found\\\"}))\\n\\n\\n# -- Lead tool dispatch (14 tools) --\\nTOOL_HANDLERS = {\\n \\\"bash\\\": lambda **kw: _run_bash(kw[\\\"command\\\"]),\\n \\\"read_file\\\": lambda **kw: _run_read(kw[\\\"path\\\"], kw.get(\\\"limit\\\")),\\n \\\"write_file\\\": lambda **kw: _run_write(kw[\\\"path\\\"], kw[\\\"content\\\"]),\\n \\\"edit_file\\\": lambda **kw: _run_edit(kw[\\\"path\\\"], kw[\\\"old_text\\\"], kw[\\\"new_text\\\"]),\\n \\\"spawn_teammate\\\": lambda **kw: TEAM.spawn(kw[\\\"name\\\"], kw[\\\"role\\\"], kw[\\\"prompt\\\"]),\\n \\\"list_teammates\\\": lambda **kw: TEAM.list_all(),\\n \\\"send_message\\\": lambda **kw: BUS.send(\\\"lead\\\", kw[\\\"to\\\"], kw[\\\"content\\\"], kw.get(\\\"msg_type\\\", \\\"message\\\")),\\n \\\"read_inbox\\\": lambda **kw: json.dumps(BUS.read_inbox(\\\"lead\\\"), indent=2),\\n \\\"broadcast\\\": lambda **kw: BUS.broadcast(\\\"lead\\\", kw[\\\"content\\\"], TEAM.member_names()),\\n \\\"shutdown_request\\\": lambda **kw: handle_shutdown_request(kw[\\\"teammate\\\"]),\\n \\\"shutdown_response\\\": lambda **kw: _check_shutdown_status(kw.get(\\\"request_id\\\", \\\"\\\")),\\n \\\"plan_approval\\\": lambda **kw: handle_plan_review(kw[\\\"request_id\\\"], kw[\\\"approve\\\"], kw.get(\\\"feedback\\\", \\\"\\\")),\\n \\\"idle\\\": lambda **kw: \\\"Lead does not idle.\\\",\\n \\\"claim_task\\\": lambda **kw: claim_task(kw[\\\"task_id\\\"], \\\"lead\\\"),\\n}\\n\\n# these base tools are unchanged from s02\\nTOOLS = [\\n {\\\"name\\\": \\\"bash\\\", \\\"description\\\": \\\"Run a shell command.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"command\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"command\\\"]}},\\n {\\\"name\\\": \\\"read_file\\\", \\\"description\\\": \\\"Read file contents.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"limit\\\": {\\\"type\\\": \\\"integer\\\"}}, \\\"required\\\": [\\\"path\\\"]}},\\n {\\\"name\\\": \\\"write_file\\\", \\\"description\\\": \\\"Write content to file.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"content\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\", \\\"content\\\"]}},\\n {\\\"name\\\": \\\"edit_file\\\", \\\"description\\\": \\\"Replace exact text in file.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"path\\\": {\\\"type\\\": \\\"string\\\"}, \\\"old_text\\\": {\\\"type\\\": \\\"string\\\"}, \\\"new_text\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"path\\\", \\\"old_text\\\", \\\"new_text\\\"]}},\\n {\\\"name\\\": \\\"spawn_teammate\\\", \\\"description\\\": \\\"Spawn an autonomous teammate.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"name\\\": {\\\"type\\\": \\\"string\\\"}, \\\"role\\\": {\\\"type\\\": \\\"string\\\"}, \\\"prompt\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"name\\\", \\\"role\\\", \\\"prompt\\\"]}},\\n {\\\"name\\\": \\\"list_teammates\\\", \\\"description\\\": \\\"List all teammates.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {}}},\\n {\\\"name\\\": \\\"send_message\\\", \\\"description\\\": \\\"Send a message to a teammate.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"to\\\": {\\\"type\\\": \\\"string\\\"}, \\\"content\\\": {\\\"type\\\": \\\"string\\\"}, \\\"msg_type\\\": {\\\"type\\\": \\\"string\\\", \\\"enum\\\": list(VALID_MSG_TYPES)}}, \\\"required\\\": [\\\"to\\\", \\\"content\\\"]}},\\n {\\\"name\\\": \\\"read_inbox\\\", \\\"description\\\": \\\"Read and drain the lead's inbox.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {}}},\\n {\\\"name\\\": \\\"broadcast\\\", \\\"description\\\": \\\"Send a message to all teammates.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"content\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"content\\\"]}},\\n {\\\"name\\\": \\\"shutdown_request\\\", \\\"description\\\": \\\"Request a teammate to shut down.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"teammate\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"teammate\\\"]}},\\n {\\\"name\\\": \\\"shutdown_response\\\", \\\"description\\\": \\\"Check shutdown request status.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"request_id\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"request_id\\\"]}},\\n {\\\"name\\\": \\\"plan_approval\\\", \\\"description\\\": \\\"Approve or reject a teammate's plan.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"request_id\\\": {\\\"type\\\": \\\"string\\\"}, \\\"approve\\\": {\\\"type\\\": \\\"boolean\\\"}, \\\"feedback\\\": {\\\"type\\\": \\\"string\\\"}}, \\\"required\\\": [\\\"request_id\\\", \\\"approve\\\"]}},\\n {\\\"name\\\": \\\"idle\\\", \\\"description\\\": \\\"Enter idle state (for lead -- rarely used).\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {}}},\\n {\\\"name\\\": \\\"claim_task\\\", \\\"description\\\": \\\"Claim a task from the board by ID.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {\\\"task_id\\\": {\\\"type\\\": \\\"integer\\\"}}, \\\"required\\\": [\\\"task_id\\\"]}},\\n]\\n\\n\\ndef agent_loop(messages: list):\\n while True:\\n inbox = BUS.read_inbox(\\\"lead\\\")\\n if inbox:\\n messages.append({\\n \\\"role\\\": \\\"user\\\",\\n \\\"content\\\": f\\\"{json.dumps(inbox, indent=2)} \\\",\\n })\\n messages.append({\\n \\\"role\\\": \\\"assistant\\\",\\n \\\"content\\\": \\\"Noted inbox messages.\\\",\\n })\\n response = client.messages.create(\\n model=MODEL,\\n system=SYSTEM,\\n messages=messages,\\n tools=TOOLS,\\n max_tokens=8000,\\n )\\n messages.append({\\\"role\\\": \\\"assistant\\\", \\\"content\\\": response.content})\\n if response.stop_reason != \\\"tool_use\\\":\\n return\\n results = []\\n for block in response.content:\\n if block.type == \\\"tool_use\\\":\\n handler = TOOL_HANDLERS.get(block.name)\\n try:\\n output = handler(**block.input) if handler else f\\\"Unknown tool: {block.name}\\\"\\n except Exception as e:\\n output = f\\\"Error: {e}\\\"\\n print(f\\\"> {block.name}: {str(output)[:200]}\\\")\\n results.append({\\n \\\"type\\\": \\\"tool_result\\\",\\n \\\"tool_use_id\\\": block.id,\\n \\\"content\\\": str(output),\\n })\\n messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": results})\\n\\n\\nif __name__ == \\\"__main__\\\":\\n history = []\\n while True:\\n try:\\n query = input(\\\"\\\\033[36ms11 >> \\\\033[0m\\\")\\n except (EOFError, KeyboardInterrupt):\\n break\\n if query.strip().lower() in (\\\"q\\\", \\\"exit\\\", \\\"\\\"):\\n break\\n if query.strip() == \\\"/team\\\":\\n print(TEAM.list_all())\\n continue\\n if query.strip() == \\\"/inbox\\\":\\n print(json.dumps(BUS.read_inbox(\\\"lead\\\"), indent=2))\\n continue\\n if query.strip() == \\\"/tasks\\\":\\n TASKS_DIR.mkdir(exist_ok=True)\\n for f in sorted(TASKS_DIR.glob(\\\"task_*.json\\\")):\\n t = json.loads(f.read_text())\\n marker = {\\\"pending\\\": \\\"[ ]\\\", \\\"in_progress\\\": \\\"[>]\\\", \\\"completed\\\": \\\"[x]\\\"}.get(t[\\\"status\\\"], \\\"[?]\\\")\\n owner = f\\\" @{t['owner']}\\\" if t.get(\\\"owner\\\") else \\\"\\\"\\n print(f\\\" {marker} #{t['id']}: {t['subject']}{owner}\\\")\\n continue\\n history.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": query})\\n agent_loop(history)\\n print()\\n\"\n },\n {\n \"id\": \"s12\",\n \"filename\": \"s12_worktree_task_isolation.py\",\n \"title\": \"Worktree + Task Isolation\",\n \"subtitle\": \"Isolate by Directory\",\n \"loc\": 689,\n \"tools\": [\n \"bash\",\n \"read_file\",\n \"write_file\",\n \"edit_file\",\n \"task_create\",\n \"task_list\",\n \"task_get\",\n \"task_update\",\n \"task_bind_worktree\",\n \"worktree_create\",\n \"worktree_list\",\n \"worktree_status\",\n \"worktree_run\",\n \"worktree_remove\",\n \"worktree_keep\",\n \"worktree_events\"\n ],\n \"newTools\": [\n \"task_create\",\n \"task_list\",\n \"task_get\",\n \"task_update\",\n \"task_bind_worktree\",\n \"worktree_create\",\n \"worktree_list\",\n \"worktree_status\",\n \"worktree_run\",\n \"worktree_remove\",\n \"worktree_keep\",\n \"worktree_events\"\n ],\n \"coreAddition\": \"Composable worktree lifecycle + event stream over a shared task board\",\n \"keyInsight\": \"Each works in its own directory; tasks manage goals, worktrees manage directories, bound by ID\",\n \"classes\": [\n {\n \"name\": \"EventBus\",\n \"startLine\": 82,\n \"endLine\": 120\n },\n {\n \"name\": \"TaskManager\",\n \"startLine\": 121,\n \"endLine\": 218\n },\n {\n \"name\": \"WorktreeManager\",\n \"startLine\": 224,\n \"endLine\": 472\n }\n ],\n \"functions\": [\n {\n \"name\": \"detect_repo_root\",\n \"signature\": \"def detect_repo_root(cwd: Path)\",\n \"startLine\": 52\n },\n {\n \"name\": \"safe_path\",\n \"signature\": \"def safe_path(p: str)\",\n \"startLine\": 477\n },\n {\n \"name\": \"run_bash\",\n \"signature\": \"def run_bash(command: str)\",\n \"startLine\": 484\n },\n {\n \"name\": \"run_read\",\n \"signature\": \"def run_read(path: str, limit: int = None)\",\n \"startLine\": 503\n },\n {\n \"name\": \"run_write\",\n \"signature\": \"def run_write(path: str, content: str)\",\n \"startLine\": 513\n },\n {\n \"name\": \"run_edit\",\n \"signature\": \"def run_edit(path: str, old_text: str, new_text: str)\",\n \"startLine\": 523\n },\n {\n \"name\": \"agent_loop\",\n \"signature\": \"def agent_loop(messages: list)\",\n \"startLine\": 728\n }\n ],\n \"layer\": \"collaboration\",\n \"source\": \"#!/usr/bin/env python3\\n\\\"\\\"\\\"\\ns12_worktree_task_isolation.py - Worktree + Task Isolation\\n\\nDirectory-level isolation for parallel task execution.\\nTasks are the control plane and worktrees are the execution plane.\\n\\n .tasks/task_12.json\\n {\\n \\\"id\\\": 12,\\n \\\"subject\\\": \\\"Implement auth refactor\\\",\\n \\\"status\\\": \\\"in_progress\\\",\\n \\\"worktree\\\": \\\"auth-refactor\\\"\\n }\\n\\n .worktrees/index.json\\n {\\n \\\"worktrees\\\": [\\n {\\n \\\"name\\\": \\\"auth-refactor\\\",\\n \\\"path\\\": \\\".../.worktrees/auth-refactor\\\",\\n \\\"branch\\\": \\\"wt/auth-refactor\\\",\\n \\\"task_id\\\": 12,\\n \\\"status\\\": \\\"active\\\"\\n }\\n ]\\n }\\n\\nKey insight: \\\"Isolate by directory, coordinate by task ID.\\\"\\n\\\"\\\"\\\"\\n\\nimport json\\nimport os\\nimport re\\nimport subprocess\\nimport time\\nfrom pathlib import Path\\n\\nfrom anthropic import Anthropic\\nfrom dotenv import load_dotenv\\n\\nload_dotenv(override=True)\\n\\nif os.getenv(\\\"ANTHROPIC_BASE_URL\\\"):\\n os.environ.pop(\\\"ANTHROPIC_AUTH_TOKEN\\\", None)\\n\\nWORKDIR = Path.cwd()\\nclient = Anthropic(base_url=os.getenv(\\\"ANTHROPIC_BASE_URL\\\"))\\nMODEL = os.environ[\\\"MODEL_ID\\\"]\\n\\n\\ndef detect_repo_root(cwd: Path) -> Path | None:\\n \\\"\\\"\\\"Return git repo root if cwd is inside a repo, else None.\\\"\\\"\\\"\\n try:\\n r = subprocess.run(\\n [\\\"git\\\", \\\"rev-parse\\\", \\\"--show-toplevel\\\"],\\n cwd=cwd,\\n capture_output=True,\\n text=True,\\n timeout=10,\\n )\\n if r.returncode != 0:\\n return None\\n root = Path(r.stdout.strip())\\n return root if root.exists() else None\\n except Exception:\\n return None\\n\\n\\nREPO_ROOT = detect_repo_root(WORKDIR) or WORKDIR\\n\\nSYSTEM = (\\n f\\\"You are a coding agent at {WORKDIR}. \\\"\\n \\\"Use task + worktree tools for multi-task work. \\\"\\n \\\"For parallel or risky changes: create tasks, allocate worktree lanes, \\\"\\n \\\"run commands in those lanes, then choose keep/remove for closeout. \\\"\\n \\\"Use worktree_events when you need lifecycle visibility.\\\"\\n)\\n\\n\\n# -- EventBus: append-only lifecycle events for observability --\\nclass EventBus:\\n def __init__(self, event_log_path: Path):\\n self.path = event_log_path\\n self.path.parent.mkdir(parents=True, exist_ok=True)\\n if not self.path.exists():\\n self.path.write_text(\\\"\\\")\\n\\n def emit(\\n self,\\n event: str,\\n task: dict | None = None,\\n worktree: dict | None = None,\\n error: str | None = None,\\n ):\\n payload = {\\n \\\"event\\\": event,\\n \\\"ts\\\": time.time(),\\n \\\"task\\\": task or {},\\n \\\"worktree\\\": worktree or {},\\n }\\n if error:\\n payload[\\\"error\\\"] = error\\n with self.path.open(\\\"a\\\", encoding=\\\"utf-8\\\") as f:\\n f.write(json.dumps(payload) + \\\"\\\\n\\\")\\n\\n def list_recent(self, limit: int = 20) -> str:\\n n = max(1, min(int(limit or 20), 200))\\n lines = self.path.read_text(encoding=\\\"utf-8\\\").splitlines()\\n recent = lines[-n:]\\n items = []\\n for line in recent:\\n try:\\n items.append(json.loads(line))\\n except Exception:\\n items.append({\\\"event\\\": \\\"parse_error\\\", \\\"raw\\\": line})\\n return json.dumps(items, indent=2)\\n\\n\\n# -- TaskManager: persistent task board with optional worktree binding --\\nclass TaskManager:\\n def __init__(self, tasks_dir: Path):\\n self.dir = tasks_dir\\n self.dir.mkdir(parents=True, exist_ok=True)\\n self._next_id = self._max_id() + 1\\n\\n def _max_id(self) -> int:\\n ids = []\\n for f in self.dir.glob(\\\"task_*.json\\\"):\\n try:\\n ids.append(int(f.stem.split(\\\"_\\\")[1]))\\n except Exception:\\n pass\\n return max(ids) if ids else 0\\n\\n def _path(self, task_id: int) -> Path:\\n return self.dir / f\\\"task_{task_id}.json\\\"\\n\\n def _load(self, task_id: int) -> dict:\\n path = self._path(task_id)\\n if not path.exists():\\n raise ValueError(f\\\"Task {task_id} not found\\\")\\n return json.loads(path.read_text())\\n\\n def _save(self, task: dict):\\n self._path(task[\\\"id\\\"]).write_text(json.dumps(task, indent=2))\\n\\n def create(self, subject: str, description: str = \\\"\\\") -> str:\\n task = {\\n \\\"id\\\": self._next_id,\\n \\\"subject\\\": subject,\\n \\\"description\\\": description,\\n \\\"status\\\": \\\"pending\\\",\\n \\\"owner\\\": \\\"\\\",\\n \\\"worktree\\\": \\\"\\\",\\n \\\"blockedBy\\\": [],\\n \\\"created_at\\\": time.time(),\\n \\\"updated_at\\\": time.time(),\\n }\\n self._save(task)\\n self._next_id += 1\\n return json.dumps(task, indent=2)\\n\\n def get(self, task_id: int) -> str:\\n return json.dumps(self._load(task_id), indent=2)\\n\\n def exists(self, task_id: int) -> bool:\\n return self._path(task_id).exists()\\n\\n def update(self, task_id: int, status: str = None, owner: str = None) -> str:\\n task = self._load(task_id)\\n if status:\\n if status not in (\\\"pending\\\", \\\"in_progress\\\", \\\"completed\\\"):\\n raise ValueError(f\\\"Invalid status: {status}\\\")\\n task[\\\"status\\\"] = status\\n if owner is not None:\\n task[\\\"owner\\\"] = owner\\n task[\\\"updated_at\\\"] = time.time()\\n self._save(task)\\n return json.dumps(task, indent=2)\\n\\n def bind_worktree(self, task_id: int, worktree: str, owner: str = \\\"\\\") -> str:\\n task = self._load(task_id)\\n task[\\\"worktree\\\"] = worktree\\n if owner:\\n task[\\\"owner\\\"] = owner\\n if task[\\\"status\\\"] == \\\"pending\\\":\\n task[\\\"status\\\"] = \\\"in_progress\\\"\\n task[\\\"updated_at\\\"] = time.time()\\n self._save(task)\\n return json.dumps(task, indent=2)\\n\\n def unbind_worktree(self, task_id: int) -> str:\\n task = self._load(task_id)\\n task[\\\"worktree\\\"] = \\\"\\\"\\n task[\\\"updated_at\\\"] = time.time()\\n self._save(task)\\n return json.dumps(task, indent=2)\\n\\n def list_all(self) -> str:\\n tasks = []\\n for f in sorted(self.dir.glob(\\\"task_*.json\\\")):\\n tasks.append(json.loads(f.read_text()))\\n if not tasks:\\n return \\\"No tasks.\\\"\\n lines = []\\n for t in tasks:\\n marker = {\\n \\\"pending\\\": \\\"[ ]\\\",\\n \\\"in_progress\\\": \\\"[>]\\\",\\n \\\"completed\\\": \\\"[x]\\\",\\n }.get(t[\\\"status\\\"], \\\"[?]\\\")\\n owner = f\\\" owner={t['owner']}\\\" if t.get(\\\"owner\\\") else \\\"\\\"\\n wt = f\\\" wt={t['worktree']}\\\" if t.get(\\\"worktree\\\") else \\\"\\\"\\n lines.append(f\\\"{marker} #{t['id']}: {t['subject']}{owner}{wt}\\\")\\n return \\\"\\\\n\\\".join(lines)\\n\\n\\nTASKS = TaskManager(REPO_ROOT / \\\".tasks\\\")\\nEVENTS = EventBus(REPO_ROOT / \\\".worktrees\\\" / \\\"events.jsonl\\\")\\n\\n\\n# -- WorktreeManager: create/list/run/remove git worktrees + lifecycle index --\\nclass WorktreeManager:\\n def __init__(self, repo_root: Path, tasks: TaskManager, events: EventBus):\\n self.repo_root = repo_root\\n self.tasks = tasks\\n self.events = events\\n self.dir = repo_root / \\\".worktrees\\\"\\n self.dir.mkdir(parents=True, exist_ok=True)\\n self.index_path = self.dir / \\\"index.json\\\"\\n if not self.index_path.exists():\\n self.index_path.write_text(json.dumps({\\\"worktrees\\\": []}, indent=2))\\n self.git_available = self._is_git_repo()\\n\\n def _is_git_repo(self) -> bool:\\n try:\\n r = subprocess.run(\\n [\\\"git\\\", \\\"rev-parse\\\", \\\"--is-inside-work-tree\\\"],\\n cwd=self.repo_root,\\n capture_output=True,\\n text=True,\\n timeout=10,\\n )\\n return r.returncode == 0\\n except Exception:\\n return False\\n\\n def _run_git(self, args: list[str]) -> str:\\n if not self.git_available:\\n raise RuntimeError(\\\"Not in a git repository. worktree tools require git.\\\")\\n r = subprocess.run(\\n [\\\"git\\\", *args],\\n cwd=self.repo_root,\\n capture_output=True,\\n text=True,\\n timeout=120,\\n )\\n if r.returncode != 0:\\n msg = (r.stdout + r.stderr).strip()\\n raise RuntimeError(msg or f\\\"git {' '.join(args)} failed\\\")\\n return (r.stdout + r.stderr).strip() or \\\"(no output)\\\"\\n\\n def _load_index(self) -> dict:\\n return json.loads(self.index_path.read_text())\\n\\n def _save_index(self, data: dict):\\n self.index_path.write_text(json.dumps(data, indent=2))\\n\\n def _find(self, name: str) -> dict | None:\\n idx = self._load_index()\\n for wt in idx.get(\\\"worktrees\\\", []):\\n if wt.get(\\\"name\\\") == name:\\n return wt\\n return None\\n\\n def _validate_name(self, name: str):\\n if not re.fullmatch(r\\\"[A-Za-z0-9._-]{1,40}\\\", name or \\\"\\\"):\\n raise ValueError(\\n \\\"Invalid worktree name. Use 1-40 chars: letters, numbers, ., _, -\\\"\\n )\\n\\n def create(self, name: str, task_id: int = None, base_ref: str = \\\"HEAD\\\") -> str:\\n self._validate_name(name)\\n if self._find(name):\\n raise ValueError(f\\\"Worktree '{name}' already exists in index\\\")\\n if task_id is not None and not self.tasks.exists(task_id):\\n raise ValueError(f\\\"Task {task_id} not found\\\")\\n\\n path = self.dir / name\\n branch = f\\\"wt/{name}\\\"\\n self.events.emit(\\n \\\"worktree.create.before\\\",\\n task={\\\"id\\\": task_id} if task_id is not None else {},\\n worktree={\\\"name\\\": name, \\\"base_ref\\\": base_ref},\\n )\\n try:\\n self._run_git([\\\"worktree\\\", \\\"add\\\", \\\"-b\\\", branch, str(path), base_ref])\\n\\n entry = {\\n \\\"name\\\": name,\\n \\\"path\\\": str(path),\\n \\\"branch\\\": branch,\\n \\\"task_id\\\": task_id,\\n \\\"status\\\": \\\"active\\\",\\n \\\"created_at\\\": time.time(),\\n }\\n\\n idx = self._load_index()\\n idx[\\\"worktrees\\\"].append(entry)\\n self._save_index(idx)\\n\\n if task_id is not None:\\n self.tasks.bind_worktree(task_id, name)\\n\\n self.events.emit(\\n \\\"worktree.create.after\\\",\\n task={\\\"id\\\": task_id} if task_id is not None else {},\\n worktree={\\n \\\"name\\\": name,\\n \\\"path\\\": str(path),\\n \\\"branch\\\": branch,\\n \\\"status\\\": \\\"active\\\",\\n },\\n )\\n return json.dumps(entry, indent=2)\\n except Exception as e:\\n self.events.emit(\\n \\\"worktree.create.failed\\\",\\n task={\\\"id\\\": task_id} if task_id is not None else {},\\n worktree={\\\"name\\\": name, \\\"base_ref\\\": base_ref},\\n error=str(e),\\n )\\n raise\\n\\n def list_all(self) -> str:\\n idx = self._load_index()\\n wts = idx.get(\\\"worktrees\\\", [])\\n if not wts:\\n return \\\"No worktrees in index.\\\"\\n lines = []\\n for wt in wts:\\n suffix = f\\\" task={wt['task_id']}\\\" if wt.get(\\\"task_id\\\") else \\\"\\\"\\n lines.append(\\n f\\\"[{wt.get('status', 'unknown')}] {wt['name']} -> \\\"\\n f\\\"{wt['path']} ({wt.get('branch', '-')}){suffix}\\\"\\n )\\n return \\\"\\\\n\\\".join(lines)\\n\\n def status(self, name: str) -> str:\\n wt = self._find(name)\\n if not wt:\\n return f\\\"Error: Unknown worktree '{name}'\\\"\\n path = Path(wt[\\\"path\\\"])\\n if not path.exists():\\n return f\\\"Error: Worktree path missing: {path}\\\"\\n r = subprocess.run(\\n [\\\"git\\\", \\\"status\\\", \\\"--short\\\", \\\"--branch\\\"],\\n cwd=path,\\n capture_output=True,\\n text=True,\\n timeout=60,\\n )\\n text = (r.stdout + r.stderr).strip()\\n return text or \\\"Clean worktree\\\"\\n\\n def run(self, name: str, command: str) -> str:\\n dangerous = [\\\"rm -rf /\\\", \\\"sudo\\\", \\\"shutdown\\\", \\\"reboot\\\", \\\"> /dev/\\\"]\\n if any(d in command for d in dangerous):\\n return \\\"Error: Dangerous command blocked\\\"\\n\\n wt = self._find(name)\\n if not wt:\\n return f\\\"Error: Unknown worktree '{name}'\\\"\\n path = Path(wt[\\\"path\\\"])\\n if not path.exists():\\n return f\\\"Error: Worktree path missing: {path}\\\"\\n\\n try:\\n r = subprocess.run(\\n command,\\n shell=True,\\n cwd=path,\\n capture_output=True,\\n text=True,\\n timeout=300,\\n )\\n out = (r.stdout + r.stderr).strip()\\n return out[:50000] if out else \\\"(no output)\\\"\\n except subprocess.TimeoutExpired:\\n return \\\"Error: Timeout (300s)\\\"\\n\\n def remove(self, name: str, force: bool = False, complete_task: bool = False) -> str:\\n wt = self._find(name)\\n if not wt:\\n return f\\\"Error: Unknown worktree '{name}'\\\"\\n\\n self.events.emit(\\n \\\"worktree.remove.before\\\",\\n task={\\\"id\\\": wt.get(\\\"task_id\\\")} if wt.get(\\\"task_id\\\") is not None else {},\\n worktree={\\\"name\\\": name, \\\"path\\\": wt.get(\\\"path\\\")},\\n )\\n try:\\n args = [\\\"worktree\\\", \\\"remove\\\"]\\n if force:\\n args.append(\\\"--force\\\")\\n args.append(wt[\\\"path\\\"])\\n self._run_git(args)\\n\\n if complete_task and wt.get(\\\"task_id\\\") is not None:\\n task_id = wt[\\\"task_id\\\"]\\n before = json.loads(self.tasks.get(task_id))\\n self.tasks.update(task_id, status=\\\"completed\\\")\\n self.tasks.unbind_worktree(task_id)\\n self.events.emit(\\n \\\"task.completed\\\",\\n task={\\n \\\"id\\\": task_id,\\n \\\"subject\\\": before.get(\\\"subject\\\", \\\"\\\"),\\n \\\"status\\\": \\\"completed\\\",\\n },\\n worktree={\\\"name\\\": name},\\n )\\n\\n idx = self._load_index()\\n for item in idx.get(\\\"worktrees\\\", []):\\n if item.get(\\\"name\\\") == name:\\n item[\\\"status\\\"] = \\\"removed\\\"\\n item[\\\"removed_at\\\"] = time.time()\\n self._save_index(idx)\\n\\n self.events.emit(\\n \\\"worktree.remove.after\\\",\\n task={\\\"id\\\": wt.get(\\\"task_id\\\")} if wt.get(\\\"task_id\\\") is not None else {},\\n worktree={\\\"name\\\": name, \\\"path\\\": wt.get(\\\"path\\\"), \\\"status\\\": \\\"removed\\\"},\\n )\\n return f\\\"Removed worktree '{name}'\\\"\\n except Exception as e:\\n self.events.emit(\\n \\\"worktree.remove.failed\\\",\\n task={\\\"id\\\": wt.get(\\\"task_id\\\")} if wt.get(\\\"task_id\\\") is not None else {},\\n worktree={\\\"name\\\": name, \\\"path\\\": wt.get(\\\"path\\\")},\\n error=str(e),\\n )\\n raise\\n\\n def keep(self, name: str) -> str:\\n wt = self._find(name)\\n if not wt:\\n return f\\\"Error: Unknown worktree '{name}'\\\"\\n\\n idx = self._load_index()\\n kept = None\\n for item in idx.get(\\\"worktrees\\\", []):\\n if item.get(\\\"name\\\") == name:\\n item[\\\"status\\\"] = \\\"kept\\\"\\n item[\\\"kept_at\\\"] = time.time()\\n kept = item\\n self._save_index(idx)\\n\\n self.events.emit(\\n \\\"worktree.keep\\\",\\n task={\\\"id\\\": wt.get(\\\"task_id\\\")} if wt.get(\\\"task_id\\\") is not None else {},\\n worktree={\\n \\\"name\\\": name,\\n \\\"path\\\": wt.get(\\\"path\\\"),\\n \\\"status\\\": \\\"kept\\\",\\n },\\n )\\n return json.dumps(kept, indent=2) if kept else f\\\"Error: Unknown worktree '{name}'\\\"\\n\\n\\nWORKTREES = WorktreeManager(REPO_ROOT, TASKS, EVENTS)\\n\\n\\n# -- Base tools (kept minimal, same style as previous sessions) --\\ndef safe_path(p: str) -> Path:\\n path = (WORKDIR / p).resolve()\\n if not path.is_relative_to(WORKDIR):\\n raise ValueError(f\\\"Path escapes workspace: {p}\\\")\\n return path\\n\\n\\ndef run_bash(command: str) -> str:\\n dangerous = [\\\"rm -rf /\\\", \\\"sudo\\\", \\\"shutdown\\\", \\\"reboot\\\", \\\"> /dev/\\\"]\\n if any(d in command for d in dangerous):\\n return \\\"Error: Dangerous command blocked\\\"\\n try:\\n r = subprocess.run(\\n command,\\n shell=True,\\n cwd=WORKDIR,\\n capture_output=True,\\n text=True,\\n timeout=120,\\n )\\n out = (r.stdout + r.stderr).strip()\\n return out[:50000] if out else \\\"(no output)\\\"\\n except subprocess.TimeoutExpired:\\n return \\\"Error: Timeout (120s)\\\"\\n\\n\\ndef run_read(path: str, limit: int = None) -> str:\\n try:\\n lines = safe_path(path).read_text().splitlines()\\n if limit and limit < len(lines):\\n lines = lines[:limit] + [f\\\"... ({len(lines) - limit} more)\\\"]\\n return \\\"\\\\n\\\".join(lines)[:50000]\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\n\\ndef run_write(path: str, content: str) -> str:\\n try:\\n fp = safe_path(path)\\n fp.parent.mkdir(parents=True, exist_ok=True)\\n fp.write_text(content)\\n return f\\\"Wrote {len(content)} bytes\\\"\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\n\\ndef run_edit(path: str, old_text: str, new_text: str) -> str:\\n try:\\n fp = safe_path(path)\\n c = fp.read_text()\\n if old_text not in c:\\n return f\\\"Error: Text not found in {path}\\\"\\n fp.write_text(c.replace(old_text, new_text, 1))\\n return f\\\"Edited {path}\\\"\\n except Exception as e:\\n return f\\\"Error: {e}\\\"\\n\\n\\nTOOL_HANDLERS = {\\n \\\"bash\\\": lambda **kw: run_bash(kw[\\\"command\\\"]),\\n \\\"read_file\\\": lambda **kw: run_read(kw[\\\"path\\\"], kw.get(\\\"limit\\\")),\\n \\\"write_file\\\": lambda **kw: run_write(kw[\\\"path\\\"], kw[\\\"content\\\"]),\\n \\\"edit_file\\\": lambda **kw: run_edit(kw[\\\"path\\\"], kw[\\\"old_text\\\"], kw[\\\"new_text\\\"]),\\n \\\"task_create\\\": lambda **kw: TASKS.create(kw[\\\"subject\\\"], kw.get(\\\"description\\\", \\\"\\\")),\\n \\\"task_list\\\": lambda **kw: TASKS.list_all(),\\n \\\"task_get\\\": lambda **kw: TASKS.get(kw[\\\"task_id\\\"]),\\n \\\"task_update\\\": lambda **kw: TASKS.update(kw[\\\"task_id\\\"], kw.get(\\\"status\\\"), kw.get(\\\"owner\\\")),\\n \\\"task_bind_worktree\\\": lambda **kw: TASKS.bind_worktree(kw[\\\"task_id\\\"], kw[\\\"worktree\\\"], kw.get(\\\"owner\\\", \\\"\\\")),\\n \\\"worktree_create\\\": lambda **kw: WORKTREES.create(kw[\\\"name\\\"], kw.get(\\\"task_id\\\"), kw.get(\\\"base_ref\\\", \\\"HEAD\\\")),\\n \\\"worktree_list\\\": lambda **kw: WORKTREES.list_all(),\\n \\\"worktree_status\\\": lambda **kw: WORKTREES.status(kw[\\\"name\\\"]),\\n \\\"worktree_run\\\": lambda **kw: WORKTREES.run(kw[\\\"name\\\"], kw[\\\"command\\\"]),\\n \\\"worktree_keep\\\": lambda **kw: WORKTREES.keep(kw[\\\"name\\\"]),\\n \\\"worktree_remove\\\": lambda **kw: WORKTREES.remove(kw[\\\"name\\\"], kw.get(\\\"force\\\", False), kw.get(\\\"complete_task\\\", False)),\\n \\\"worktree_events\\\": lambda **kw: EVENTS.list_recent(kw.get(\\\"limit\\\", 20)),\\n}\\n\\nTOOLS = [\\n {\\n \\\"name\\\": \\\"bash\\\",\\n \\\"description\\\": \\\"Run a shell command in the current workspace (blocking).\\\",\\n \\\"input_schema\\\": {\\n \\\"type\\\": \\\"object\\\",\\n \\\"properties\\\": {\\\"command\\\": {\\\"type\\\": \\\"string\\\"}},\\n \\\"required\\\": [\\\"command\\\"],\\n },\\n },\\n {\\n \\\"name\\\": \\\"read_file\\\",\\n \\\"description\\\": \\\"Read file contents.\\\",\\n \\\"input_schema\\\": {\\n \\\"type\\\": \\\"object\\\",\\n \\\"properties\\\": {\\n \\\"path\\\": {\\\"type\\\": \\\"string\\\"},\\n \\\"limit\\\": {\\\"type\\\": \\\"integer\\\"},\\n },\\n \\\"required\\\": [\\\"path\\\"],\\n },\\n },\\n {\\n \\\"name\\\": \\\"write_file\\\",\\n \\\"description\\\": \\\"Write content to file.\\\",\\n \\\"input_schema\\\": {\\n \\\"type\\\": \\\"object\\\",\\n \\\"properties\\\": {\\n \\\"path\\\": {\\\"type\\\": \\\"string\\\"},\\n \\\"content\\\": {\\\"type\\\": \\\"string\\\"},\\n },\\n \\\"required\\\": [\\\"path\\\", \\\"content\\\"],\\n },\\n },\\n {\\n \\\"name\\\": \\\"edit_file\\\",\\n \\\"description\\\": \\\"Replace exact text in file.\\\",\\n \\\"input_schema\\\": {\\n \\\"type\\\": \\\"object\\\",\\n \\\"properties\\\": {\\n \\\"path\\\": {\\\"type\\\": \\\"string\\\"},\\n \\\"old_text\\\": {\\\"type\\\": \\\"string\\\"},\\n \\\"new_text\\\": {\\\"type\\\": \\\"string\\\"},\\n },\\n \\\"required\\\": [\\\"path\\\", \\\"old_text\\\", \\\"new_text\\\"],\\n },\\n },\\n {\\n \\\"name\\\": \\\"task_create\\\",\\n \\\"description\\\": \\\"Create a new task on the shared task board.\\\",\\n \\\"input_schema\\\": {\\n \\\"type\\\": \\\"object\\\",\\n \\\"properties\\\": {\\n \\\"subject\\\": {\\\"type\\\": \\\"string\\\"},\\n \\\"description\\\": {\\\"type\\\": \\\"string\\\"},\\n },\\n \\\"required\\\": [\\\"subject\\\"],\\n },\\n },\\n {\\n \\\"name\\\": \\\"task_list\\\",\\n \\\"description\\\": \\\"List all tasks with status, owner, and worktree binding.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {}},\\n },\\n {\\n \\\"name\\\": \\\"task_get\\\",\\n \\\"description\\\": \\\"Get task details by ID.\\\",\\n \\\"input_schema\\\": {\\n \\\"type\\\": \\\"object\\\",\\n \\\"properties\\\": {\\\"task_id\\\": {\\\"type\\\": \\\"integer\\\"}},\\n \\\"required\\\": [\\\"task_id\\\"],\\n },\\n },\\n {\\n \\\"name\\\": \\\"task_update\\\",\\n \\\"description\\\": \\\"Update task status or owner.\\\",\\n \\\"input_schema\\\": {\\n \\\"type\\\": \\\"object\\\",\\n \\\"properties\\\": {\\n \\\"task_id\\\": {\\\"type\\\": \\\"integer\\\"},\\n \\\"status\\\": {\\n \\\"type\\\": \\\"string\\\",\\n \\\"enum\\\": [\\\"pending\\\", \\\"in_progress\\\", \\\"completed\\\"],\\n },\\n \\\"owner\\\": {\\\"type\\\": \\\"string\\\"},\\n },\\n \\\"required\\\": [\\\"task_id\\\"],\\n },\\n },\\n {\\n \\\"name\\\": \\\"task_bind_worktree\\\",\\n \\\"description\\\": \\\"Bind a task to a worktree name.\\\",\\n \\\"input_schema\\\": {\\n \\\"type\\\": \\\"object\\\",\\n \\\"properties\\\": {\\n \\\"task_id\\\": {\\\"type\\\": \\\"integer\\\"},\\n \\\"worktree\\\": {\\\"type\\\": \\\"string\\\"},\\n \\\"owner\\\": {\\\"type\\\": \\\"string\\\"},\\n },\\n \\\"required\\\": [\\\"task_id\\\", \\\"worktree\\\"],\\n },\\n },\\n {\\n \\\"name\\\": \\\"worktree_create\\\",\\n \\\"description\\\": \\\"Create a git worktree and optionally bind it to a task.\\\",\\n \\\"input_schema\\\": {\\n \\\"type\\\": \\\"object\\\",\\n \\\"properties\\\": {\\n \\\"name\\\": {\\\"type\\\": \\\"string\\\"},\\n \\\"task_id\\\": {\\\"type\\\": \\\"integer\\\"},\\n \\\"base_ref\\\": {\\\"type\\\": \\\"string\\\"},\\n },\\n \\\"required\\\": [\\\"name\\\"],\\n },\\n },\\n {\\n \\\"name\\\": \\\"worktree_list\\\",\\n \\\"description\\\": \\\"List worktrees tracked in .worktrees/index.json.\\\",\\n \\\"input_schema\\\": {\\\"type\\\": \\\"object\\\", \\\"properties\\\": {}},\\n },\\n {\\n \\\"name\\\": \\\"worktree_status\\\",\\n \\\"description\\\": \\\"Show git status for one worktree.\\\",\\n \\\"input_schema\\\": {\\n \\\"type\\\": \\\"object\\\",\\n \\\"properties\\\": {\\\"name\\\": {\\\"type\\\": \\\"string\\\"}},\\n \\\"required\\\": [\\\"name\\\"],\\n },\\n },\\n {\\n \\\"name\\\": \\\"worktree_run\\\",\\n \\\"description\\\": \\\"Run a shell command in a named worktree directory.\\\",\\n \\\"input_schema\\\": {\\n \\\"type\\\": \\\"object\\\",\\n \\\"properties\\\": {\\n \\\"name\\\": {\\\"type\\\": \\\"string\\\"},\\n \\\"command\\\": {\\\"type\\\": \\\"string\\\"},\\n },\\n \\\"required\\\": [\\\"name\\\", \\\"command\\\"],\\n },\\n },\\n {\\n \\\"name\\\": \\\"worktree_remove\\\",\\n \\\"description\\\": \\\"Remove a worktree and optionally mark its bound task completed.\\\",\\n \\\"input_schema\\\": {\\n \\\"type\\\": \\\"object\\\",\\n \\\"properties\\\": {\\n \\\"name\\\": {\\\"type\\\": \\\"string\\\"},\\n \\\"force\\\": {\\\"type\\\": \\\"boolean\\\"},\\n \\\"complete_task\\\": {\\\"type\\\": \\\"boolean\\\"},\\n },\\n \\\"required\\\": [\\\"name\\\"],\\n },\\n },\\n {\\n \\\"name\\\": \\\"worktree_keep\\\",\\n \\\"description\\\": \\\"Mark a worktree as kept in lifecycle state without removing it.\\\",\\n \\\"input_schema\\\": {\\n \\\"type\\\": \\\"object\\\",\\n \\\"properties\\\": {\\\"name\\\": {\\\"type\\\": \\\"string\\\"}},\\n \\\"required\\\": [\\\"name\\\"],\\n },\\n },\\n {\\n \\\"name\\\": \\\"worktree_events\\\",\\n \\\"description\\\": \\\"List recent worktree/task lifecycle events from .worktrees/events.jsonl.\\\",\\n \\\"input_schema\\\": {\\n \\\"type\\\": \\\"object\\\",\\n \\\"properties\\\": {\\\"limit\\\": {\\\"type\\\": \\\"integer\\\"}},\\n },\\n },\\n]\\n\\n\\ndef agent_loop(messages: list):\\n while True:\\n response = client.messages.create(\\n model=MODEL,\\n system=SYSTEM,\\n messages=messages,\\n tools=TOOLS,\\n max_tokens=8000,\\n )\\n messages.append({\\\"role\\\": \\\"assistant\\\", \\\"content\\\": response.content})\\n if response.stop_reason != \\\"tool_use\\\":\\n return\\n\\n results = []\\n for block in response.content:\\n if block.type == \\\"tool_use\\\":\\n handler = TOOL_HANDLERS.get(block.name)\\n try:\\n output = handler(**block.input) if handler else f\\\"Unknown tool: {block.name}\\\"\\n except Exception as e:\\n output = f\\\"Error: {e}\\\"\\n print(f\\\"> {block.name}: {str(output)[:200]}\\\")\\n results.append(\\n {\\n \\\"type\\\": \\\"tool_result\\\",\\n \\\"tool_use_id\\\": block.id,\\n \\\"content\\\": str(output),\\n }\\n )\\n messages.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": results})\\n\\n\\nif __name__ == \\\"__main__\\\":\\n print(f\\\"Repo root for s12: {REPO_ROOT}\\\")\\n if not WORKTREES.git_available:\\n print(\\\"Note: Not in a git repo. worktree_* tools will return errors.\\\")\\n\\n history = []\\n while True:\\n try:\\n query = input(\\\"\\\\033[36ms12 >> \\\\033[0m\\\")\\n except (EOFError, KeyboardInterrupt):\\n break\\n if query.strip().lower() in (\\\"q\\\", \\\"exit\\\", \\\"\\\"):\\n break\\n history.append({\\\"role\\\": \\\"user\\\", \\\"content\\\": query})\\n agent_loop(history)\\n print()\\n\"\n }\n ],\n \"diffs\": [\n {\n \"from\": \"s01\",\n \"to\": \"s02\",\n \"newClasses\": [],\n \"newFunctions\": [\n \"safe_path\",\n \"run_read\",\n \"run_write\",\n \"run_edit\"\n ],\n \"newTools\": [\n \"read_file\",\n \"write_file\",\n \"edit_file\"\n ],\n \"locDelta\": 31\n },\n {\n \"from\": \"s02\",\n \"to\": \"s03\",\n \"newClasses\": [\n \"TodoManager\"\n ],\n \"newFunctions\": [],\n \"newTools\": [\n \"todo\"\n ],\n \"locDelta\": 56\n },\n {\n \"from\": \"s03\",\n \"to\": \"s04\",\n \"newClasses\": [],\n \"newFunctions\": [\n \"run_subagent\"\n ],\n \"newTools\": [\n \"task\"\n ],\n \"locDelta\": -25\n },\n {\n \"from\": \"s04\",\n \"to\": \"s05\",\n \"newClasses\": [\n \"SkillLoader\"\n ],\n \"newFunctions\": [],\n \"newTools\": [\n \"load_skill\"\n ],\n \"locDelta\": 36\n },\n {\n \"from\": \"s05\",\n \"to\": \"s06\",\n \"newClasses\": [],\n \"newFunctions\": [\n \"estimate_tokens\",\n \"micro_compact\",\n \"auto_compact\"\n ],\n \"newTools\": [\n \"compact\"\n ],\n \"locDelta\": 18\n },\n {\n \"from\": \"s06\",\n \"to\": \"s07\",\n \"newClasses\": [\n \"TaskManager\"\n ],\n \"newFunctions\": [],\n \"newTools\": [\n \"task_create\",\n \"task_update\",\n \"task_list\",\n \"task_get\"\n ],\n \"locDelta\": 2\n },\n {\n \"from\": \"s07\",\n \"to\": \"s08\",\n \"newClasses\": [\n \"BackgroundManager\"\n ],\n \"newFunctions\": [],\n \"newTools\": [\n \"background_run\",\n \"check_background\"\n ],\n \"locDelta\": -9\n },\n {\n \"from\": \"s08\",\n \"to\": \"s09\",\n \"newClasses\": [\n \"MessageBus\",\n \"TeammateManager\"\n ],\n \"newFunctions\": [\n \"_safe_path\",\n \"_run_bash\",\n \"_run_read\",\n \"_run_write\",\n \"_run_edit\"\n ],\n \"newTools\": [\n \"alice\",\n \"send_message\",\n \"read_inbox\",\n \"spawn_teammate\",\n \"list_teammates\",\n \"broadcast\"\n ],\n \"locDelta\": 150\n },\n {\n \"from\": \"s09\",\n \"to\": \"s10\",\n \"newClasses\": [],\n \"newFunctions\": [\n \"handle_shutdown_request\",\n \"handle_plan_review\",\n \"_check_shutdown_status\"\n ],\n \"newTools\": [\n \"shutdown_response\",\n \"plan_approval\",\n \"shutdown_request\"\n ],\n \"locDelta\": 71\n },\n {\n \"from\": \"s10\",\n \"to\": \"s11\",\n \"newClasses\": [],\n \"newFunctions\": [\n \"scan_unclaimed_tasks\",\n \"claim_task\",\n \"make_identity_block\"\n ],\n \"newTools\": [\n \"idle\",\n \"claim_task\"\n ],\n \"locDelta\": 80\n },\n {\n \"from\": \"s11\",\n \"to\": \"s12\",\n \"newClasses\": [\n \"EventBus\",\n \"TaskManager\",\n \"WorktreeManager\"\n ],\n \"newFunctions\": [\n \"detect_repo_root\",\n \"safe_path\",\n \"run_bash\",\n \"run_read\",\n \"run_write\",\n \"run_edit\"\n ],\n \"newTools\": [\n \"task_create\",\n \"task_list\",\n \"task_get\",\n \"task_update\",\n \"task_bind_worktree\",\n \"worktree_create\",\n \"worktree_list\",\n \"worktree_status\",\n \"worktree_run\",\n \"worktree_remove\",\n \"worktree_keep\",\n \"worktree_events\"\n ],\n \"locDelta\": 195\n }\n ]\n}"
},
{
"path": "web/src/data/scenarios/s01.json",
"content": "{\n \"version\": \"s01\",\n \"title\": \"The Agent Loop\",\n \"description\": \"A minimal agent that uses only bash to accomplish tasks\",\n \"steps\": [\n {\n \"type\": \"user_message\",\n \"content\": \"Create a file called hello.py that prints 'Hello, World!'\",\n \"annotation\": \"User sends a task to the agent\"\n },\n {\n \"type\": \"assistant_text\",\n \"content\": \"I'll create that file for you using bash.\",\n \"annotation\": \"The model decides to use its only tool: bash\"\n },\n {\n \"type\": \"tool_call\",\n \"content\": \"echo 'print(\\\"Hello, World!\\\")' > hello.py\",\n \"toolName\": \"bash\",\n \"annotation\": \"Tool call: the model generates a bash command\"\n },\n {\n \"type\": \"tool_result\",\n \"content\": \"\",\n \"toolName\": \"bash\",\n \"annotation\": \"Bash returns empty output (success)\"\n },\n {\n \"type\": \"assistant_text\",\n \"content\": \"Let me verify the file was created correctly.\",\n \"annotation\": \"The loop continues: model sees the result and decides next action\"\n },\n {\n \"type\": \"tool_call\",\n \"content\": \"cat hello.py\",\n \"toolName\": \"bash\",\n \"annotation\": \"Another bash call to verify\"\n },\n {\n \"type\": \"tool_result\",\n \"content\": \"print(\\\"Hello, World!\\\")\",\n \"toolName\": \"bash\",\n \"annotation\": \"File contents confirmed\"\n },\n {\n \"type\": \"assistant_text\",\n \"content\": \"Done! I've created hello.py with the print statement. You can run it with `python hello.py`.\",\n \"annotation\": \"stop_reason != 'tool_use' -> loop breaks, agent responds to user\"\n }\n ]\n}\n"
},
{
"path": "web/src/data/scenarios/s02.json",
"content": "{\n \"version\": \"s02\",\n \"title\": \"Tools\",\n \"description\": \"Agent with read, write, edit, and bash tools\",\n \"steps\": [\n {\n \"type\": \"user_message\",\n \"content\": \"Add a function called greet(name) to hello.py\",\n \"annotation\": \"User requests a code modification\"\n },\n {\n \"type\": \"assistant_text\",\n \"content\": \"Let me read the current file first.\",\n \"annotation\": \"Model chooses read_file instead of bash cat\"\n },\n {\n \"type\": \"tool_call\",\n \"content\": \"hello.py\",\n \"toolName\": \"read_file\",\n \"annotation\": \"Dedicated read tool provides cleaner file access\"\n },\n {\n \"type\": \"tool_result\",\n \"content\": \"print(\\\"Hello, World!\\\")\",\n \"toolName\": \"read_file\",\n \"annotation\": \"File contents returned to the model\"\n },\n {\n \"type\": \"tool_call\",\n \"content\": \"hello.py\\n---\\ndef greet(name):\\n print(f\\\"Hello, {name}!\\\")\\n\\ngreet(\\\"World\\\")\",\n \"toolName\": \"write_file\",\n \"annotation\": \"write_file replaces the entire file content\"\n },\n {\n \"type\": \"tool_result\",\n \"content\": \"File written successfully\",\n \"toolName\": \"write_file\",\n \"annotation\": \"Write confirmed\"\n },\n {\n \"type\": \"assistant_text\",\n \"content\": \"I've updated hello.py with a greet(name) function that takes a name parameter.\",\n \"annotation\": \"4 tools beat 1: each tool has a clear, focused purpose\"\n }\n ]\n}\n"
},
{
"path": "web/src/data/scenarios/s03.json",
"content": "{\n \"version\": \"s03\",\n \"title\": \"TodoWrite\",\n \"description\": \"Agent creates a visible plan before executing\",\n \"steps\": [\n {\n \"type\": \"user_message\",\n \"content\": \"Build a CLI calculator that supports add, subtract, multiply, divide\",\n \"annotation\": \"A multi-step task that benefits from planning\"\n },\n {\n \"type\": \"tool_call\",\n \"content\": \"[\\n { \\\"id\\\": 1, \\\"task\\\": \\\"Create calc.py with argument parsing\\\", \\\"status\\\": \\\"in_progress\\\" },\\n { \\\"id\\\": 2, \\\"task\\\": \\\"Implement 4 operations\\\", \\\"status\\\": \\\"pending\\\" },\\n { \\\"id\\\": 3, \\\"task\\\": \\\"Add error handling\\\", \\\"status\\\": \\\"pending\\\" },\\n { \\\"id\\\": 4, \\\"task\\\": \\\"Test all operations\\\", \\\"status\\\": \\\"pending\\\" }\\n]\",\n \"toolName\": \"todo_write\",\n \"annotation\": \"TodoManager creates a visible plan BEFORE any code is written\"\n },\n {\n \"type\": \"tool_result\",\n \"content\": \"Todo list created with 4 tasks\",\n \"toolName\": \"todo_write\",\n \"annotation\": \"The plan is now visible and trackable\"\n },\n {\n \"type\": \"tool_call\",\n \"content\": \"calc.py\\n---\\nimport sys\\n\\ndef calculate(op, a, b):\\n ops = {'+': a+b, '-': a-b, '*': a*b, '/': a/b}\\n return ops.get(op, 'Unknown operator')\\n\\nif __name__ == '__main__':\\n a, op, b = float(sys.argv[1]), sys.argv[2], float(sys.argv[3])\\n print(calculate(op, a, b))\",\n \"toolName\": \"write_file\",\n \"annotation\": \"Executing task #1 and #2: create file with operations\"\n },\n {\n \"type\": \"tool_result\",\n \"content\": \"File written successfully\",\n \"toolName\": \"write_file\",\n \"annotation\": \"Code written\"\n },\n {\n \"type\": \"tool_call\",\n \"content\": \"python calc.py 10 + 5\",\n \"toolName\": \"bash\",\n \"annotation\": \"Executing task #4: testing\"\n },\n {\n \"type\": \"tool_result\",\n \"content\": \"15.0\",\n \"toolName\": \"bash\",\n \"annotation\": \"Test passes\"\n },\n {\n \"type\": \"assistant_text\",\n \"content\": \"Calculator built and tested. All 4 operations work correctly.\",\n \"annotation\": \"Visible plans improve task completion by making progress trackable\"\n }\n ]\n}\n"
},
{
"path": "web/src/data/scenarios/s04.json",
"content": "{\n \"version\": \"s04\",\n \"title\": \"Subagents\",\n \"description\": \"Agent spawns a subagent for isolated subtask execution\",\n \"steps\": [\n {\n \"type\": \"user_message\",\n \"content\": \"Refactor the calculator: split into modules and add tests\",\n \"annotation\": \"Complex task suitable for delegation\"\n },\n {\n \"type\": \"assistant_text\",\n \"content\": \"I'll delegate the test writing to a subagent while I handle the refactoring.\",\n \"annotation\": \"Agent decides to split work using the Task tool\"\n },\n {\n \"type\": \"tool_call\",\n \"content\": \"Write unit tests for calc.py covering all 4 operations and edge cases like division by zero\",\n \"toolName\": \"task\",\n \"annotation\": \"Task tool spawns a subagent with its own isolated context\"\n },\n {\n \"type\": \"system_event\",\n \"content\": \"[Subagent spawned] Context: fresh messages[], same tools, scoped prompt\",\n \"annotation\": \"Subagent gets clean context -- no confusion from parent's conversation\"\n },\n {\n \"type\": \"tool_result\",\n \"content\": \"Created test_calc.py with 8 test cases. All passing.\",\n \"toolName\": \"task\",\n \"annotation\": \"Subagent completes and returns a summary to parent\"\n },\n {\n \"type\": \"tool_call\",\n \"content\": \"calc.py -> calc/operations.py, calc/cli.py\",\n \"toolName\": \"write_file\",\n \"annotation\": \"Parent continues with its own task: refactoring\"\n },\n {\n \"type\": \"tool_result\",\n \"content\": \"Files written successfully\",\n \"toolName\": \"write_file\",\n \"annotation\": \"Modules split\"\n },\n {\n \"type\": \"assistant_text\",\n \"content\": \"Refactored into calc/operations.py and calc/cli.py. Tests were written by a subagent and are all passing.\",\n \"annotation\": \"Context isolation prevents confusion between tasks\"\n }\n ]\n}\n"
},
{
"path": "web/src/data/scenarios/s05.json",
"content": "{\n \"version\": \"s05\",\n \"title\": \"Skills\",\n \"description\": \"Agent loads external knowledge from SKILL.md files\",\n \"steps\": [\n {\n \"type\": \"user_message\",\n \"content\": \"Set up a new Next.js project with TypeScript\",\n \"annotation\": \"Task requires framework-specific knowledge\"\n },\n {\n \"type\": \"system_event\",\n \"content\": \"[SkillLoader] Scanning for SKILL.md files... Found: nextjs.SKILL.md\",\n \"annotation\": \"SkillLoader finds a matching skill file before the loop runs\"\n },\n {\n \"type\": \"system_event\",\n \"content\": \"[Skill injected] nextjs.SKILL.md -> tool_result message appended\",\n \"annotation\": \"Skill content is injected as a tool_result, not bloating the system prompt\"\n },\n {\n \"type\": \"assistant_text\",\n \"content\": \"I have knowledge about Next.js setup. Let me create the project structure.\",\n \"annotation\": \"Model now has framework-specific knowledge from the skill file\"\n },\n {\n \"type\": \"tool_call\",\n \"content\": \"npx create-next-app@latest my-app --typescript --tailwind --eslint --app --src-dir\",\n \"toolName\": \"bash\",\n \"annotation\": \"Using skill-provided knowledge for correct flags\"\n },\n {\n \"type\": \"tool_result\",\n \"content\": \"Success! Created my-app at /workspace/my-app\",\n \"toolName\": \"bash\",\n \"annotation\": \"Project created with correct configuration\"\n },\n {\n \"type\": \"assistant_text\",\n \"content\": \"Next.js project created with TypeScript, Tailwind CSS, ESLint, App Router, and src/ directory.\",\n \"annotation\": \"Skills inject via tool_result, not system prompt -- keeping context lean\"\n }\n ]\n}\n"
},
{
"path": "web/src/data/scenarios/s06.json",
"content": "{\n \"version\": \"s06\",\n \"title\": \"Compact\",\n \"description\": \"Agent compresses context when tokens exceed threshold\",\n \"steps\": [\n {\n \"type\": \"user_message\",\n \"content\": \"Analyze all Python files in this project and summarize each one\",\n \"annotation\": \"A task that will generate large amounts of context\"\n },\n {\n \"type\": \"tool_call\",\n \"content\": \"find . -name '*.py' -type f\",\n \"toolName\": \"bash\",\n \"annotation\": \"Finding all files to analyze\"\n },\n {\n \"type\": \"tool_result\",\n \"content\": \"./src/main.py\\n./src/utils.py\\n./src/models.py\\n./tests/test_main.py\\n... (28 more files)\",\n \"toolName\": \"bash\",\n \"annotation\": \"32 files found -- reading all of them will generate massive context\"\n },\n {\n \"type\": \"assistant_text\",\n \"content\": \"I'll read and analyze each file systematically.\",\n \"annotation\": \"Agent begins processing files one by one\"\n },\n {\n \"type\": \"tool_call\",\n \"content\": \"./src/main.py\",\n \"toolName\": \"read_file\",\n \"annotation\": \"Reading file 1 of 32...\"\n },\n {\n \"type\": \"tool_result\",\n \"content\": \"[... 450 lines of code ...]\",\n \"toolName\": \"read_file\",\n \"annotation\": \"Large file content added to messages[]\"\n },\n {\n \"type\": \"system_event\",\n \"content\": \"[ContextManager] Token count: 148,231 / 150,000 threshold\\nTriggering 3-layer compression:\\n 1. Drop tool_result content > 1000 chars -> replaced with summary\\n 2. Merge sequential assistant turns\\n 3. Summarize oldest 50% of conversation\\nResult: 148,231 -> 42,087 tokens\",\n \"annotation\": \"Compression fires automatically when approaching the limit\"\n },\n {\n \"type\": \"assistant_text\",\n \"content\": \"Continuing analysis with compressed context. Files 1-18 have been summarized, proceeding with remaining files.\",\n \"annotation\": \"Forgetting old tool results enables infinite work -- the agent can process any number of files\"\n }\n ]\n}\n"
},
{
"path": "web/src/data/scenarios/s07.json",
"content": "{\n \"version\": \"s07\",\n \"title\": \"Tasks\",\n \"description\": \"Agent uses persistent file-based tasks with dependencies\",\n \"steps\": [\n {\n \"type\": \"user_message\",\n \"content\": \"Build a REST API with user auth, database, and deployment config\",\n \"annotation\": \"Multi-part task requiring dependency tracking\"\n },\n {\n \"type\": \"tool_call\",\n \"content\": \"create_task(\\\"Set up database schema\\\", deps=[])\\ncreate_task(\\\"Implement user model\\\", deps=[1])\\ncreate_task(\\\"Add auth endpoints\\\", deps=[2])\\ncreate_task(\\\"Write deployment config\\\", deps=[3])\",\n \"toolName\": \"task_manager\",\n \"annotation\": \"TaskManager creates file-based tasks with explicit dependencies\"\n },\n {\n \"type\": \"tool_result\",\n \"content\": \"Created 4 tasks in .tasks/\\n #1 Set up database schema [ready]\\n #2 Implement user model [blocked by #1]\\n #3 Add auth endpoints [blocked by #2]\\n #4 Write deployment config [blocked by #3]\",\n \"toolName\": \"task_manager\",\n \"annotation\": \"Tasks persisted to .tasks/ directory as JSON files\"\n },\n {\n \"type\": \"tool_call\",\n \"content\": \"schema.sql\\n---\\nCREATE TABLE users (id SERIAL PRIMARY KEY, email TEXT UNIQUE, password_hash TEXT);\",\n \"toolName\": \"write_file\",\n \"annotation\": \"Working on task #1\"\n },\n {\n \"type\": \"tool_result\",\n \"content\": \"File written successfully\",\n \"toolName\": \"write_file\",\n \"annotation\": \"Schema created\"\n },\n {\n \"type\": \"tool_call\",\n \"content\": \"complete_task(1)\",\n \"toolName\": \"task_manager\",\n \"annotation\": \"Completing task #1 automatically unblocks task #2\"\n },\n {\n \"type\": \"tool_result\",\n \"content\": \"Task #1 completed. Task #2 is now ready.\",\n \"toolName\": \"task_manager\",\n \"annotation\": \"File-based persistence outlives process memory\"\n },\n {\n \"type\": \"assistant_text\",\n \"content\": \"Database schema is done. Moving to user model implementation.\",\n \"annotation\": \"Dependency graph ensures tasks execute in correct order\"\n }\n ]\n}\n"
},
{
"path": "web/src/data/scenarios/s08.json",
"content": "{\n \"version\": \"s08\",\n \"title\": \"Background Tasks\",\n \"description\": \"Agent executes long-running tasks in background threads\",\n \"steps\": [\n {\n \"type\": \"user_message\",\n \"content\": \"Run the test suite and while it's running, fix the linting errors\",\n \"annotation\": \"Two tasks that can run concurrently\"\n },\n {\n \"type\": \"assistant_text\",\n \"content\": \"I'll start the tests in the background and fix linting in parallel.\",\n \"annotation\": \"Agent recognizes opportunity for concurrent execution\"\n },\n {\n \"type\": \"tool_call\",\n \"content\": \"pytest tests/ -v --timeout=300\",\n \"toolName\": \"background_task\",\n \"annotation\": \"BackgroundManager spawns a thread for long-running test suite\"\n },\n {\n \"type\": \"system_event\",\n \"content\": \"[BackgroundManager] Thread spawned: task_001\\nRunning: pytest tests/ -v --timeout=300\\nStatus: in_progress\",\n \"annotation\": \"Thread is running independently -- main loop continues\"\n },\n {\n \"type\": \"tool_result\",\n \"content\": \"Background task started: task_001\",\n \"toolName\": \"background_task\",\n \"annotation\": \"Control returns immediately to the agent\"\n },\n {\n \"type\": \"tool_call\",\n \"content\": \"ruff check --fix .\",\n \"toolName\": \"bash\",\n \"annotation\": \"Agent works on linting while tests run in background\"\n },\n {\n \"type\": \"tool_result\",\n \"content\": \"Fixed 12 linting errors in 5 files\",\n \"toolName\": \"bash\",\n \"annotation\": \"Linting fixed concurrently with tests\"\n },\n {\n \"type\": \"system_event\",\n \"content\": \"[Notification] Background task_001 completed\\nResult: 47 tests passed, 2 failed\\nFailed: test_auth.py::test_token_expiry, test_api.py::test_rate_limit\",\n \"annotation\": \"Background thread notifies main loop via notification queue\"\n },\n {\n \"type\": \"assistant_text\",\n \"content\": \"Linting is fixed. Tests finished: 47 passed, 2 failed. I'll fix the failing tests now.\",\n \"annotation\": \"Non-blocking execution via threads + queue enables parallel work\"\n }\n ]\n}\n"
},
{
"path": "web/src/data/scenarios/s09.json",
"content": "{\n \"version\": \"s09\",\n \"title\": \"Agent Teams\",\n \"description\": \"Persistent teammates communicate via JSONL inboxes\",\n \"steps\": [\n {\n \"type\": \"user_message\",\n \"content\": \"Build a full-stack app: frontend in React, backend in Python, with tests\",\n \"annotation\": \"Large task requiring team collaboration\"\n },\n {\n \"type\": \"tool_call\",\n \"content\": \"spawn_teammate(\\n name: \\\"frontend-dev\\\",\\n role: \\\"React specialist\\\"\\n)\\nspawn_teammate(\\n name: \\\"backend-dev\\\",\\n role: \\\"Python API specialist\\\"\\n)\",\n \"toolName\": \"teammate_manager\",\n \"annotation\": \"Unlike subagents (s04) that die after one task, teammates persist\"\n },\n {\n \"type\": \"system_event\",\n \"content\": \"[TeammateManager] Team created:\\n .team/config.json written\\n frontend-dev [idle] - own thread, own agent loop\\n backend-dev [idle] - own thread, own agent loop\",\n \"annotation\": \"Each teammate runs its own while(tool_use) loop in a separate thread\"\n },\n {\n \"type\": \"tool_call\",\n \"content\": \"send_message(\\n to: \\\"backend-dev\\\",\\n content: \\\"Build the user API: GET/POST /users\\\"\\n)\",\n \"toolName\": \"send_message\",\n \"annotation\": \"Message appended to .team/inbox/backend-dev.jsonl\"\n },\n {\n \"type\": \"system_event\",\n \"content\": \"[backend-dev] Checking inbox... 1 new message\\n From: lead\\n Content: \\\"Build the user API: GET/POST /users\\\"\\n (inbox drained after read)\",\n \"annotation\": \"Inbox is read-then-clear: prevents duplicate processing\"\n },\n {\n \"type\": \"system_event\",\n \"content\": \"[backend-dev -> frontend-dev] Reply via inbox:\\n \\\"API ready: GET /users, POST /users, GET /users/:id\\\"\\n Written to .team/inbox/frontend-dev.jsonl\",\n \"annotation\": \"Teammates message each other through the same JSONL mechanism\"\n },\n {\n \"type\": \"assistant_text\",\n \"content\": \"Team is communicating via file-based inboxes. Each teammate has its own .jsonl file -- append to send, drain to read.\",\n \"annotation\": \"JSONL inboxes decouple sender timing from receiver timing\"\n }\n ]\n}\n"
},
{
"path": "web/src/data/scenarios/s10.json",
"content": "{\n \"version\": \"s10\",\n \"title\": \"Team Protocols\",\n \"description\": \"Request-response protocols for shutdown and plan approval\",\n \"steps\": [\n {\n \"type\": \"user_message\",\n \"content\": \"Work is done. Shut down the team gracefully.\",\n \"annotation\": \"Can't just kill threads -- need coordinated shutdown\"\n },\n {\n \"type\": \"tool_call\",\n \"content\": \"shutdown_request(\\n teammate: \\\"backend-dev\\\",\\n request_id: \\\"req_abc123\\\"\\n)\",\n \"toolName\": \"shutdown_teammate\",\n \"annotation\": \"request_id correlates the request with the response\"\n },\n {\n \"type\": \"system_event\",\n \"content\": \"[Shutdown FSM] req_abc123: pending\\n Lead -> backend-dev inbox:\\n { type: \\\"shutdown_request\\\", request_id: \\\"req_abc123\\\" }\",\n \"annotation\": \"FSM tracks state: pending -> approved | rejected\"\n },\n {\n \"type\": \"system_event\",\n \"content\": \"[backend-dev] Received shutdown request req_abc123\\n Current task: none (idle)\\n Decision: approve\\n -> lead inbox: { type: \\\"shutdown_response\\\", request_id: \\\"req_abc123\\\", approve: true }\",\n \"annotation\": \"Teammate decides whether to approve based on its own state\"\n },\n {\n \"type\": \"system_event\",\n \"content\": \"[Shutdown FSM] req_abc123: approved\\n backend-dev thread stopped\\n Status: idle -> shutdown\",\n \"annotation\": \"Same request_id pattern works for plan approval too\"\n },\n {\n \"type\": \"assistant_text\",\n \"content\": \"backend-dev shut down gracefully. The same request_id + FSM pattern handles plan approval: teammate submits plan, lead approves/rejects with the correlated request_id.\",\n \"annotation\": \"One pattern (request_id correlation), two applications (shutdown + plan approval)\"\n }\n ]\n}\n"
},
{
"path": "web/src/data/scenarios/s11.json",
"content": "{\n \"version\": \"s11\",\n \"title\": \"Autonomous Agents\",\n \"description\": \"Teammates self-govern with idle cycles and auto-claiming\",\n \"steps\": [\n {\n \"type\": \"user_message\",\n \"content\": \"Implement the full feature backlog\",\n \"annotation\": \"Open-ended task for autonomous team execution\"\n },\n {\n \"type\": \"tool_call\",\n \"content\": \"create_tasks([\\n { title: \\\"User authentication\\\", status: \\\"open\\\" },\\n { title: \\\"Dashboard UI\\\", status: \\\"open\\\" },\\n { title: \\\"API rate limiting\\\", status: \\\"open\\\" },\\n { title: \\\"Integration tests\\\", status: \\\"open\\\" }\\n])\",\n \"toolName\": \"task_manager\",\n \"annotation\": \"Tasks created with no owner -- available for auto-claiming\"\n },\n {\n \"type\": \"tool_result\",\n \"content\": \"4 tasks created on shared board\",\n \"toolName\": \"task_manager\",\n \"annotation\": \"Unassigned tasks visible to all teammates\"\n },\n {\n \"type\": \"system_event\",\n \"content\": \"[frontend-dev] Idle cycle triggered\\n 1. Check inbox -> 0 messages\\n 2. Poll task board -> found unowned task #2 \\\"Dashboard UI\\\"\\n 3. Auto-claim task #2\\n 4. Status: idle -> working\",\n \"annotation\": \"Idle cycle: check inbox, poll tasks, auto-claim, resume work\"\n },\n {\n \"type\": \"system_event\",\n \"content\": \"[backend-dev] Idle cycle triggered\\n 1. Check inbox -> 0 messages\\n 2. Poll task board -> found unowned task #1 \\\"User authentication\\\"\\n 3. Auto-claim task #1\\n 4. Status: idle -> working\",\n \"annotation\": \"Multiple teammates claim different tasks concurrently\"\n },\n {\n \"type\": \"system_event\",\n \"content\": \"[tester] Idle cycle triggered\\n 1. Check inbox -> 0 messages\\n 2. Poll task board -> task #4 blocked by #1, #2, #3\\n 3. No claimable tasks\\n 4. Status: idle (will retry in 30s)\",\n \"annotation\": \"Timeout-based polling prevents busy-waiting\"\n },\n {\n \"type\": \"assistant_text\",\n \"content\": \"Team is self-organizing: frontend-dev claimed Dashboard UI, backend-dev claimed User auth. Tester is waiting for dependencies to clear.\",\n \"annotation\": \"Polling + timeout makes teammates autonomous -- no micromanagement needed\"\n }\n ]\n}\n"
},
{
"path": "web/src/data/scenarios/s12.json",
"content": "{\n \"version\": \"s12\",\n \"title\": \"Worktree + Task Isolation\",\n \"description\": \"Use a shared task board with optional worktree lanes for clean parallel execution\",\n \"steps\": [\n {\n \"type\": \"user_message\",\n \"content\": \"Implement auth refactor and login UI updates in parallel\",\n \"annotation\": \"Two active tasks in one workspace would collide\"\n },\n {\n \"type\": \"tool_call\",\n \"content\": \"task_create(subject: \\\"Auth refactor\\\")\\ntask_create(subject: \\\"Login UI polish\\\")\",\n \"toolName\": \"task_manager\",\n \"annotation\": \"Shared board remains the coordination source of truth\"\n },\n {\n \"type\": \"tool_call\",\n \"content\": \"worktree_create(name: \\\"auth-refactor\\\", task_id: 1)\\nworktree_create(name: \\\"ui-login\\\")\\ntask_bind_worktree(task_id: 2, worktree: \\\"ui-login\\\")\",\n \"toolName\": \"worktree_manager\",\n \"annotation\": \"Lane allocation and task association are composable; task 2 binds after lane creation\"\n },\n {\n \"type\": \"system_event\",\n \"content\": \"worktree.create.before/after emitted\\n.tasks/task_1.json -> { status: \\\"in_progress\\\", worktree: \\\"auth-refactor\\\" }\\n.tasks/task_2.json -> { status: \\\"in_progress\\\", worktree: \\\"ui-login\\\" }\\n.worktrees/index.json updated\",\n \"annotation\": \"Control-plane state remains canonical; hook-style consumers can react to lifecycle events without owning canonical state writes\"\n },\n {\n \"type\": \"tool_call\",\n \"content\": \"worktree_run(name: \\\"auth-refactor\\\", command: \\\"pytest tests/auth -q\\\")\\nworktree_run(name: \\\"ui-login\\\", command: \\\"npm test -- login\\\")\",\n \"toolName\": \"worktree_run\",\n \"annotation\": \"In this teaching runtime, commands route by lane-scoped cwd; other runtimes may use session-level directory switches. The invariant is explicit execution context.\"\n },\n {\n \"type\": \"tool_call\",\n \"content\": \"worktree_keep(name: \\\"ui-login\\\")\\nworktree_remove(name: \\\"auth-refactor\\\", complete_task: true)\\nworktree_events(limit: 10)\",\n \"toolName\": \"worktree_manager\",\n \"annotation\": \"Closeout is explicit tool-driven state transition: mix keep/remove decisions and query lifecycle events in one pass\"\n },\n {\n \"type\": \"system_event\",\n \"content\": \"worktree.keep emitted for ui-login\\nworktree.remove.before/after emitted for auth-refactor\\ntask.completed emitted for #1\\n.worktrees/events.jsonl appended\",\n \"annotation\": \"Lifecycle transitions become explicit records while task/worktree files remain source-of-truth\"\n },\n {\n \"type\": \"assistant_text\",\n \"content\": \"Task board handles coordination, worktrees handle isolation. Parallel tracks stay clean and auditable.\",\n \"annotation\": \"Coordinate in one board, isolate by lane only where needed, and run optional policy/audit side effects from lifecycle events\"\n }\n ]\n}\n"
},
{
"path": "web/src/hooks/useDarkMode.ts",
"content": "\"use client\";\n\nimport { useState, useEffect } from \"react\";\n\nexport function useDarkMode(): boolean {\n const [isDark, setIsDark] = useState(false);\n\n useEffect(() => {\n const html = document.documentElement;\n setIsDark(html.classList.contains(\"dark\"));\n\n const observer = new MutationObserver(() => {\n setIsDark(html.classList.contains(\"dark\"));\n });\n\n observer.observe(html, { attributes: true, attributeFilter: [\"class\"] });\n return () => observer.disconnect();\n }, []);\n\n return isDark;\n}\n\nexport interface SvgPalette {\n nodeFill: string;\n nodeStroke: string;\n nodeText: string;\n activeNodeFill: string;\n activeNodeStroke: string;\n activeNodeText: string;\n endNodeFill: string;\n endNodeStroke: string;\n edgeStroke: string;\n activeEdgeStroke: string;\n arrowFill: string;\n labelFill: string;\n bgSubtle: string;\n}\n\nexport function useSvgPalette(): SvgPalette {\n const isDark = useDarkMode();\n\n if (isDark) {\n return {\n nodeFill: \"#27272a\",\n nodeStroke: \"#3f3f46\",\n nodeText: \"#d4d4d8\",\n activeNodeFill: \"#3b82f6\",\n activeNodeStroke: \"#2563eb\",\n activeNodeText: \"#ffffff\",\n endNodeFill: \"#a855f7\",\n endNodeStroke: \"#9333ea\",\n edgeStroke: \"#52525b\",\n activeEdgeStroke: \"#3b82f6\",\n arrowFill: \"#71717a\",\n labelFill: \"#a1a1aa\",\n bgSubtle: \"#18181b\",\n };\n }\n\n return {\n nodeFill: \"#e2e8f0\",\n nodeStroke: \"#cbd5e1\",\n nodeText: \"#475569\",\n activeNodeFill: \"#3b82f6\",\n activeNodeStroke: \"#2563eb\",\n activeNodeText: \"#ffffff\",\n endNodeFill: \"#a855f7\",\n endNodeStroke: \"#9333ea\",\n edgeStroke: \"#cbd5e1\",\n activeEdgeStroke: \"#3b82f6\",\n arrowFill: \"#94a3b8\",\n labelFill: \"#94a3b8\",\n bgSubtle: \"#f8fafc\",\n };\n}\n"
},
{
"path": "web/src/hooks/useSimulator.ts",
"content": "\"use client\";\n\nimport { useState, useCallback, useRef, useEffect } from \"react\";\nimport type { SimStep } from \"@/types/agent-data\";\n\ninterface SimulatorState {\n currentIndex: number;\n isPlaying: boolean;\n speed: number;\n}\n\nexport function useSimulator(steps: SimStep[]) {\n const [state, setState] = useState({\n currentIndex: -1,\n isPlaying: false,\n speed: 1,\n });\n const timerRef = useRef | null>(null);\n\n const clearTimer = useCallback(() => {\n if (timerRef.current) {\n clearTimeout(timerRef.current);\n timerRef.current = null;\n }\n }, []);\n\n const stepForward = useCallback(() => {\n setState((prev) => {\n if (prev.currentIndex >= steps.length - 1) {\n return { ...prev, isPlaying: false };\n }\n return { ...prev, currentIndex: prev.currentIndex + 1 };\n });\n }, [steps.length]);\n\n const play = useCallback(() => {\n setState((prev) => {\n if (prev.currentIndex >= steps.length - 1) {\n return prev;\n }\n return { ...prev, isPlaying: true };\n });\n }, [steps.length]);\n\n const pause = useCallback(() => {\n clearTimer();\n setState((prev) => ({ ...prev, isPlaying: false }));\n }, [clearTimer]);\n\n const reset = useCallback(() => {\n clearTimer();\n setState({ currentIndex: -1, isPlaying: false, speed: state.speed });\n }, [clearTimer, state.speed]);\n\n const setSpeed = useCallback((speed: number) => {\n setState((prev) => ({ ...prev, speed }));\n }, []);\n\n useEffect(() => {\n if (state.isPlaying && state.currentIndex < steps.length - 1) {\n const delay = 1200 / state.speed;\n timerRef.current = setTimeout(() => {\n stepForward();\n }, delay);\n } else if (state.isPlaying && state.currentIndex >= steps.length - 1) {\n setState((prev) => ({ ...prev, isPlaying: false }));\n }\n return () => clearTimer();\n }, [state.isPlaying, state.currentIndex, state.speed, steps.length, stepForward, clearTimer]);\n\n return {\n currentIndex: state.currentIndex,\n isPlaying: state.isPlaying,\n speed: state.speed,\n visibleSteps: steps.slice(0, state.currentIndex + 1),\n totalSteps: steps.length,\n isComplete: state.currentIndex >= steps.length - 1,\n play,\n pause,\n stepForward,\n reset,\n setSpeed,\n };\n}\n"
},
{
"path": "web/src/hooks/useSteppedVisualization.ts",
"content": "\"use client\";\n\nimport { useState, useCallback, useEffect, useRef } from \"react\";\n\ninterface SteppedVisualizationOptions {\n totalSteps: number;\n autoPlayInterval?: number; // ms, default 2000\n}\n\ninterface SteppedVisualizationReturn {\n currentStep: number;\n totalSteps: number;\n next: () => void;\n prev: () => void;\n reset: () => void;\n goToStep: (step: number) => void;\n isPlaying: boolean;\n toggleAutoPlay: () => void;\n isFirstStep: boolean;\n isLastStep: boolean;\n}\n\nexport function useSteppedVisualization({\n totalSteps,\n autoPlayInterval = 2000,\n}: SteppedVisualizationOptions): SteppedVisualizationReturn {\n const [currentStep, setCurrentStep] = useState(0);\n const [isPlaying, setIsPlaying] = useState(false);\n const intervalRef = useRef | null>(null);\n\n const next = useCallback(() => {\n setCurrentStep((prev) => Math.min(prev + 1, totalSteps - 1));\n }, [totalSteps]);\n\n const prev = useCallback(() => {\n setCurrentStep((prev) => Math.max(prev - 1, 0));\n }, []);\n\n const reset = useCallback(() => {\n setCurrentStep(0);\n setIsPlaying(false);\n }, []);\n\n const goToStep = useCallback(\n (step: number) => {\n setCurrentStep(Math.max(0, Math.min(step, totalSteps - 1)));\n },\n [totalSteps]\n );\n\n const toggleAutoPlay = useCallback(() => {\n setIsPlaying((prev) => !prev);\n }, []);\n\n useEffect(() => {\n if (isPlaying) {\n intervalRef.current = setInterval(() => {\n setCurrentStep((prev) => {\n if (prev >= totalSteps - 1) {\n setIsPlaying(false);\n return prev;\n }\n return prev + 1;\n });\n }, autoPlayInterval);\n }\n return () => {\n if (intervalRef.current) clearInterval(intervalRef.current);\n };\n }, [isPlaying, totalSteps, autoPlayInterval]);\n\n return {\n currentStep,\n totalSteps,\n next,\n prev,\n reset,\n goToStep,\n isPlaying,\n toggleAutoPlay,\n isFirstStep: currentStep === 0,\n isLastStep: currentStep === totalSteps - 1,\n };\n}\n"
},
{
"path": "web/src/i18n/messages/en.json",
"content": "{\n \"meta\": { \"title\": \"Learn Claude Code\", \"description\": \"Build a nano Claude Code-like agent from 0 to 1, one mechanism at a time\" },\n \"nav\": { \"home\": \"Home\", \"timeline\": \"Timeline\", \"compare\": \"Compare\", \"layers\": \"Layers\", \"github\": \"GitHub\" },\n \"home\": { \"hero_title\": \"Learn Claude Code\", \"hero_subtitle\": \"Build a nano Claude Code-like agent from 0 to 1, one mechanism at a time\", \"start\": \"Start Learning\", \"core_pattern\": \"The Core Pattern\", \"core_pattern_desc\": \"Every AI coding agent shares the same loop: call the model, execute tools, feed results back. Production systems add policy, permissions, and lifecycle layers on top.\", \"learning_path\": \"Learning Path\", \"learning_path_desc\": \"12 progressive sessions, from a simple loop to isolated autonomous execution\", \"layers_title\": \"Architectural Layers\", \"layers_desc\": \"Five orthogonal concerns that compose into a complete agent\", \"loc\": \"LOC\", \"learn_more\": \"Learn More\", \"versions_in_layer\": \"versions\", \"message_flow\": \"Message Growth\", \"message_flow_desc\": \"Watch the messages array grow as the agent loop executes\" },\n \"version\": { \"loc\": \"lines of code\", \"tools\": \"tools\", \"new\": \"New\", \"prev\": \"Previous\", \"next\": \"Next\", \"view_source\": \"View Source\", \"view_diff\": \"View Diff\", \"design_decisions\": \"Design Decisions\", \"whats_new\": \"What's New\", \"tutorial\": \"Tutorial\", \"simulator\": \"Agent Loop Simulator\", \"execution_flow\": \"Execution Flow\", \"architecture\": \"Architecture\", \"concept_viz\": \"Concept Visualization\", \"alternatives\": \"Alternatives Considered\", \"tab_learn\": \"Learn\", \"tab_simulate\": \"Simulate\", \"tab_code\": \"Code\", \"tab_deep_dive\": \"Deep Dive\" },\n \"sim\": { \"play\": \"Play\", \"pause\": \"Pause\", \"step\": \"Step\", \"reset\": \"Reset\", \"speed\": \"Speed\", \"step_of\": \"of\" },\n \"timeline\": { \"title\": \"Learning Path\", \"subtitle\": \"s01 to s12: Progressive Agent Design\", \"layer_legend\": \"Layer Legend\", \"loc_growth\": \"LOC Growth\", \"learn_more\": \"Learn More\" },\n \"layers\": {\n \"title\": \"Architectural Layers\",\n \"subtitle\": \"Five orthogonal concerns that compose into a complete agent\",\n \"tools\": \"What the agent CAN do. The foundation: tools give the model capabilities to interact with the world.\",\n \"planning\": \"How work is organized. From simple todo lists to dependency-aware task boards shared across agents.\",\n \"memory\": \"Keeping context within limits. Compression strategies that let agents work infinitely without losing coherence.\",\n \"concurrency\": \"Non-blocking execution. Background threads and notification buses for parallel work.\",\n \"collaboration\": \"Multi-agent coordination. Teams, messaging, and autonomous teammates that think for themselves.\"\n },\n \"compare\": {\n \"title\": \"Compare Versions\",\n \"subtitle\": \"See what changed between any two versions\",\n \"select_a\": \"Version A\",\n \"select_b\": \"Version B\",\n \"loc_delta\": \"LOC Delta\",\n \"lines\": \"lines\",\n \"new_tools_in_b\": \"New Tools in B\",\n \"new_classes_in_b\": \"New Classes in B\",\n \"new_functions_in_b\": \"New Functions in B\",\n \"tool_comparison\": \"Tool Comparison\",\n \"only_in\": \"Only in\",\n \"shared\": \"Shared\",\n \"none\": \"None\",\n \"source_diff\": \"Source Code Diff\",\n \"empty_hint\": \"Select two versions above to compare them.\",\n \"architecture\": \"Architecture\"\n },\n \"diff\": {\n \"new_classes\": \"New Classes\",\n \"new_tools\": \"New Tools\",\n \"new_functions\": \"New Functions\",\n \"loc_delta\": \"LOC Delta\"\n },\n \"sessions\": {\n \"s01\": \"The Agent Loop\",\n \"s02\": \"Tools\",\n \"s03\": \"TodoWrite\",\n \"s04\": \"Subagents\",\n \"s05\": \"Skills\",\n \"s06\": \"Compact\",\n \"s07\": \"Tasks\",\n \"s08\": \"Background Tasks\",\n \"s09\": \"Agent Teams\",\n \"s10\": \"Team Protocols\",\n \"s11\": \"Autonomous Agents\",\n \"s12\": \"Worktree + Task Isolation\"\n },\n \"layer_labels\": {\n \"tools\": \"Tools & Execution\",\n \"planning\": \"Planning & Coordination\",\n \"memory\": \"Memory Management\",\n \"concurrency\": \"Concurrency\",\n \"collaboration\": \"Collaboration\"\n },\n \"viz\": {\n \"s01\": \"The Agent While-Loop\",\n \"s02\": \"Tool Dispatch Map\",\n \"s03\": \"TodoWrite Nag System\",\n \"s04\": \"Subagent Context Isolation\",\n \"s05\": \"On-Demand Skill Loading\",\n \"s06\": \"Three-Layer Context Compression\",\n \"s07\": \"Task Dependency Graph\",\n \"s08\": \"Background Task Lanes\",\n \"s09\": \"Agent Team Mailboxes\",\n \"s10\": \"FSM Team Protocols\",\n \"s11\": \"Autonomous Agent Cycle\",\n \"s12\": \"Worktree Task Isolation\"\n }\n}\n"
},
{
"path": "web/src/i18n/messages/ja.json",
"content": "{\n \"meta\": { \"title\": \"Learn Claude Code\", \"description\": \"0 から 1 へ nano Claude Code-like agent を構築し、毎回 1 つの仕組みを追加\" },\n \"nav\": { \"home\": \"ホーム\", \"timeline\": \"学習パス\", \"compare\": \"バージョン比較\", \"layers\": \"アーキテクチャ層\", \"github\": \"GitHub\" },\n \"home\": { \"hero_title\": \"Learn Claude Code\", \"hero_subtitle\": \"0 から 1 へ nano Claude Code-like agent を構築し、毎回 1 つの仕組みを追加\", \"start\": \"学習を始める\", \"core_pattern\": \"コアパターン\", \"core_pattern_desc\": \"すべての AI コーディングエージェントは同じループを共有する：モデルを呼び出し、ツールを実行し、結果を返す。実運用ではこの上にポリシー、権限、ライフサイクル層が重なる。\", \"learning_path\": \"学習パス\", \"learning_path_desc\": \"12の段階的セッション、シンプルなループから分離された自律実行まで\", \"layers_title\": \"アーキテクチャ層\", \"layers_desc\": \"5つの直交する関心事が完全なエージェントを構成\", \"loc\": \"行\", \"learn_more\": \"詳細を見る\", \"versions_in_layer\": \"バージョン\", \"message_flow\": \"メッセージの増加\", \"message_flow_desc\": \"エージェントループ実行時のメッセージ配列の成長を観察\" },\n \"version\": { \"loc\": \"行のコード\", \"tools\": \"ツール\", \"new\": \"新規\", \"prev\": \"前のバージョン\", \"next\": \"次のバージョン\", \"view_source\": \"ソースを見る\", \"view_diff\": \"差分を見る\", \"design_decisions\": \"設計判断\", \"whats_new\": \"新機能\", \"tutorial\": \"チュートリアル\", \"simulator\": \"エージェントループシミュレーター\", \"execution_flow\": \"実行フロー\", \"architecture\": \"アーキテクチャ\", \"concept_viz\": \"コンセプト可視化\", \"alternatives\": \"検討された代替案\", \"tab_learn\": \"学習\", \"tab_simulate\": \"シミュレーション\", \"tab_code\": \"ソースコード\", \"tab_deep_dive\": \"詳細分析\" },\n \"sim\": { \"play\": \"再生\", \"pause\": \"一時停止\", \"step\": \"ステップ\", \"reset\": \"リセット\", \"speed\": \"速度\", \"step_of\": \"/\" },\n \"timeline\": { \"title\": \"学習パス\", \"subtitle\": \"s01からs12へ：段階的エージェント設計\", \"layer_legend\": \"レイヤー凡例\", \"loc_growth\": \"コード量の推移\", \"learn_more\": \"詳細を見る\" },\n \"layers\": {\n \"title\": \"アーキテクチャ層\",\n \"subtitle\": \"5つの直交する関心事が完全なエージェントを構成\",\n \"tools\": \"エージェントができること。基盤：ツールがモデルに外部世界と対話する能力を与える。\",\n \"planning\": \"作業の組織化。シンプルなToDoリストからエージェント間で共有される依存関係対応タスクボードまで。\",\n \"memory\": \"コンテキスト制限内での記憶保持。圧縮戦略によりエージェントが一貫性を失わずに無限に作業可能。\",\n \"concurrency\": \"ノンブロッキング実行。バックグラウンドスレッドと通知バスによる並列作業。\",\n \"collaboration\": \"マルチエージェント連携。チーム、メッセージング、自律的に考えるチームメイト。\"\n },\n \"compare\": {\n \"title\": \"バージョン比較\",\n \"subtitle\": \"任意の2つのバージョン間の変更を確認\",\n \"select_a\": \"バージョンA\",\n \"select_b\": \"バージョンB\",\n \"loc_delta\": \"コード量の差分\",\n \"lines\": \"行\",\n \"new_tools_in_b\": \"Bの新規ツール\",\n \"new_classes_in_b\": \"Bの新規クラス\",\n \"new_functions_in_b\": \"Bの新規関数\",\n \"tool_comparison\": \"ツール比較\",\n \"only_in\": \"のみ\",\n \"shared\": \"共通\",\n \"none\": \"なし\",\n \"source_diff\": \"ソースコード差分\",\n \"empty_hint\": \"上で2つのバージョンを選択して比較してください。\",\n \"architecture\": \"アーキテクチャ\"\n },\n \"diff\": {\n \"new_classes\": \"新規クラス\",\n \"new_tools\": \"新規ツール\",\n \"new_functions\": \"新規関数\",\n \"loc_delta\": \"コード量の差分\"\n },\n \"sessions\": {\n \"s01\": \"エージェントループ\",\n \"s02\": \"ツール\",\n \"s03\": \"TodoWrite\",\n \"s04\": \"サブエージェント\",\n \"s05\": \"スキル\",\n \"s06\": \"コンテキスト圧縮\",\n \"s07\": \"タスクシステム\",\n \"s08\": \"バックグラウンドタスク\",\n \"s09\": \"エージェントチーム\",\n \"s10\": \"チームプロトコル\",\n \"s11\": \"自律エージェント\",\n \"s12\": \"Worktree + タスク分離\"\n },\n \"layer_labels\": {\n \"tools\": \"ツールと実行\",\n \"planning\": \"計画と調整\",\n \"memory\": \"メモリ管理\",\n \"concurrency\": \"並行処理\",\n \"collaboration\": \"コラボレーション\"\n },\n \"viz\": {\n \"s01\": \"エージェント Whileループ\",\n \"s02\": \"ツールディスパッチマップ\",\n \"s03\": \"TodoWrite リマインドシステム\",\n \"s04\": \"サブエージェント コンテキスト分離\",\n \"s05\": \"オンデマンド スキルローディング\",\n \"s06\": \"3層コンテキスト圧縮\",\n \"s07\": \"タスク依存関係グラフ\",\n \"s08\": \"バックグラウンドタスクレーン\",\n \"s09\": \"エージェントチーム メールボックス\",\n \"s10\": \"FSM チームプロトコル\",\n \"s11\": \"自律エージェントサイクル\",\n \"s12\": \"Worktree タスク分離\"\n }\n}\n"
},
{
"path": "web/src/i18n/messages/zh.json",
"content": "{\n \"meta\": { \"title\": \"Learn Claude Code\", \"description\": \"从 0 到 1 构建 nano Claude Code-like agent，每次只加一个机制\" },\n \"nav\": { \"home\": \"首页\", \"timeline\": \"学习路径\", \"compare\": \"版本对比\", \"layers\": \"架构层\", \"github\": \"GitHub\" },\n \"home\": { \"hero_title\": \"Learn Claude Code\", \"hero_subtitle\": \"从 0 到 1 构建 nano Claude Code-like agent，每次只加一个机制\", \"start\": \"开始学习\", \"core_pattern\": \"核心模式\", \"core_pattern_desc\": \"所有 AI 编程 Agent 共享同一个循环：调用模型、执行工具、回传结果。生产级系统会在其上叠加策略、权限和生命周期层。\", \"learning_path\": \"学习路径\", \"learning_path_desc\": \"12 个渐进式课程，从简单循环到隔离化自治执行\", \"layers_title\": \"架构层次\", \"layers_desc\": \"五个正交关注点组合成完整的 Agent\", \"loc\": \"行\", \"learn_more\": \"了解更多\", \"versions_in_layer\": \"个版本\", \"message_flow\": \"消息增长\", \"message_flow_desc\": \"观察 Agent 循环执行时消息数组的增长\" },\n \"version\": { \"loc\": \"行代码\", \"tools\": \"个工具\", \"new\": \"新增\", \"prev\": \"上一版\", \"next\": \"下一版\", \"view_source\": \"查看源码\", \"view_diff\": \"查看变更\", \"design_decisions\": \"设计决策\", \"whats_new\": \"新增内容\", \"tutorial\": \"教程\", \"simulator\": \"Agent 循环模拟器\", \"execution_flow\": \"执行流程\", \"architecture\": \"架构\", \"concept_viz\": \"概念可视化\", \"alternatives\": \"替代方案\", \"tab_learn\": \"学习\", \"tab_simulate\": \"模拟\", \"tab_code\": \"源码\", \"tab_deep_dive\": \"深入探索\" },\n \"sim\": { \"play\": \"播放\", \"pause\": \"暂停\", \"step\": \"单步\", \"reset\": \"重置\", \"speed\": \"速度\", \"step_of\": \"/\" },\n \"timeline\": { \"title\": \"学习路径\", \"subtitle\": \"s01 到 s12：渐进式 Agent 设计\", \"layer_legend\": \"层次图例\", \"loc_growth\": \"代码量增长\", \"learn_more\": \"了解更多\" },\n \"layers\": {\n \"title\": \"架构层次\",\n \"subtitle\": \"五个正交关注点组合成完整的 Agent\",\n \"tools\": \"Agent 能做什么。基础层：工具赋予模型与外部世界交互的能力。\",\n \"planning\": \"如何组织工作。从简单的待办列表到跨 Agent 共享的依赖感知任务板。\",\n \"memory\": \"在上下文限制内保持记忆。压缩策略让 Agent 可以无限工作而不失去连贯性。\",\n \"concurrency\": \"非阻塞执行。后台线程和通知总线实现并行工作。\",\n \"collaboration\": \"多 Agent 协作。团队、消息传递和能独立思考的自主队友。\"\n },\n \"compare\": {\n \"title\": \"版本对比\",\n \"subtitle\": \"查看任意两个版本之间的变化\",\n \"select_a\": \"版本 A\",\n \"select_b\": \"版本 B\",\n \"loc_delta\": \"代码量差异\",\n \"lines\": \"行\",\n \"new_tools_in_b\": \"B 中新增工具\",\n \"new_classes_in_b\": \"B 中新增类\",\n \"new_functions_in_b\": \"B 中新增函数\",\n \"tool_comparison\": \"工具对比\",\n \"only_in\": \"仅在\",\n \"shared\": \"共有\",\n \"none\": \"无\",\n \"source_diff\": \"源码差异\",\n \"empty_hint\": \"请在上方选择两个版本进行对比。\",\n \"architecture\": \"架构\"\n },\n \"diff\": {\n \"new_classes\": \"新增类\",\n \"new_tools\": \"新增工具\",\n \"new_functions\": \"新增函数\",\n \"loc_delta\": \"代码量差异\"\n },\n \"sessions\": {\n \"s01\": \"Agent 循环\",\n \"s02\": \"工具\",\n \"s03\": \"TodoWrite\",\n \"s04\": \"子 Agent\",\n \"s05\": \"技能\",\n \"s06\": \"上下文压缩\",\n \"s07\": \"任务系统\",\n \"s08\": \"后台任务\",\n \"s09\": \"Agent 团队\",\n \"s10\": \"团队协议\",\n \"s11\": \"自主 Agent\",\n \"s12\": \"Worktree + 任务隔离\"\n },\n \"layer_labels\": {\n \"tools\": \"工具与执行\",\n \"planning\": \"规划与协调\",\n \"memory\": \"内存管理\",\n \"concurrency\": \"并发\",\n \"collaboration\": \"协作\"\n },\n \"viz\": {\n \"s01\": \"Agent While 循环\",\n \"s02\": \"工具分发映射\",\n \"s03\": \"TodoWrite 提醒系统\",\n \"s04\": \"子 Agent 上下文隔离\",\n \"s05\": \"按需技能加载\",\n \"s06\": \"三层上下文压缩\",\n \"s07\": \"任务依赖图\",\n \"s08\": \"后台任务通道\",\n \"s09\": \"Agent 团队邮箱\",\n \"s10\": \"FSM 团队协议\",\n \"s11\": \"自主 Agent 循环\",\n \"s12\": \"Worktree 任务隔离\"\n }\n}\n"
},
{
"path": "web/src/lib/constants.ts",
"content": "export const VERSION_ORDER = [\n \"s01\", \"s02\", \"s03\", \"s04\", \"s05\", \"s06\", \"s07\", \"s08\", \"s09\", \"s10\", \"s11\", \"s12\"\n] as const;\n\nexport const LEARNING_PATH = VERSION_ORDER;\n\nexport type VersionId = typeof LEARNING_PATH[number];\n\nexport const VERSION_META: Record = {\n s01: { title: \"The Agent Loop\", subtitle: \"Bash is All You Need\", coreAddition: \"Single-tool agent loop\", keyInsight: \"The minimal agent kernel is a while loop + one tool\", layer: \"tools\", prevVersion: null },\n s02: { title: \"Tools\", subtitle: \"One Handler Per Tool\", coreAddition: \"Tool dispatch map\", keyInsight: \"The loop stays the same; new tools register into the dispatch map\", layer: \"tools\", prevVersion: \"s01\" },\n s03: { title: \"TodoWrite\", subtitle: \"Plan Before You Act\", coreAddition: \"TodoManager + nag reminder\", keyInsight: \"An agent without a plan drifts; list the steps first, then execute\", layer: \"planning\", prevVersion: \"s02\" },\n s04: { title: \"Subagents\", subtitle: \"Clean Context Per Subtask\", coreAddition: \"Subagent spawn with isolated messages[]\", keyInsight: \"Subagents use independent messages[], keeping the main conversation clean\", layer: \"planning\", prevVersion: \"s03\" },\n s05: { title: \"Skills\", subtitle: \"Load on Demand\", coreAddition: \"SkillLoader + two-layer injection\", keyInsight: \"Inject knowledge via tool_result when needed, not upfront in the system prompt\", layer: \"planning\", prevVersion: \"s04\" },\n s06: { title: \"Compact\", subtitle: \"Three-Layer Compression\", coreAddition: \"micro-compact + auto-compact + archival\", keyInsight: \"Context will fill up; three-layer compression strategy enables infinite sessions\", layer: \"memory\", prevVersion: \"s05\" },\n s07: { title: \"Tasks\", subtitle: \"Task Graph + Dependencies\", coreAddition: \"TaskManager with file-based state + dependency graph\", keyInsight: \"A file-based task graph with ordering, parallelism, and dependencies -- the coordination backbone for multi-agent work\", layer: \"planning\", prevVersion: \"s06\" },\n s08: { title: \"Background Tasks\", subtitle: \"Background Threads + Notifications\", coreAddition: \"BackgroundManager + notification queue\", keyInsight: \"Run slow operations in the background; the agent keeps thinking ahead\", layer: \"concurrency\", prevVersion: \"s07\" },\n s09: { title: \"Agent Teams\", subtitle: \"Teammates + Mailboxes\", coreAddition: \"TeammateManager + file-based mailbox\", keyInsight: \"When one agent can't finish, delegate to persistent teammates via async mailboxes\", layer: \"collaboration\", prevVersion: \"s08\" },\n s10: { title: \"Team Protocols\", subtitle: \"Shared Communication Rules\", coreAddition: \"request_id correlation for two protocols\", keyInsight: \"One request-response pattern drives all team negotiation\", layer: \"collaboration\", prevVersion: \"s09\" },\n s11: { title: \"Autonomous Agents\", subtitle: \"Scan Board, Claim Tasks\", coreAddition: \"Task board polling + timeout-based self-governance\", keyInsight: \"Teammates scan the board and claim tasks themselves; no need for the lead to assign each one\", layer: \"collaboration\", prevVersion: \"s10\" },\n s12: { title: \"Worktree + Task Isolation\", subtitle: \"Isolate by Directory\", coreAddition: \"Composable worktree lifecycle + event stream over a shared task board\", keyInsight: \"Each works in its own directory; tasks manage goals, worktrees manage directories, bound by ID\", layer: \"collaboration\", prevVersion: \"s11\" },\n};\n\nexport const LAYERS = [\n { id: \"tools\" as const, label: \"Tools & Execution\", color: \"#3B82F6\", versions: [\"s01\", \"s02\"] },\n { id: \"planning\" as const, label: \"Planning & Coordination\", color: \"#10B981\", versions: [\"s03\", \"s04\", \"s05\", \"s07\"] },\n { id: \"memory\" as const, label: \"Memory Management\", color: \"#8B5CF6\", versions: [\"s06\"] },\n { id: \"concurrency\" as const, label: \"Concurrency\", color: \"#F59E0B\", versions: [\"s08\"] },\n { id: \"collaboration\" as const, label: \"Collaboration\", color: \"#EF4444\", versions: [\"s09\", \"s10\", \"s11\", \"s12\"] },\n] as const;\n"
},
{
"path": "web/src/lib/i18n-server.ts",
"content": "import en from \"@/i18n/messages/en.json\";\nimport zh from \"@/i18n/messages/zh.json\";\nimport ja from \"@/i18n/messages/ja.json\";\n\ntype Messages = typeof en;\n\nconst messagesMap: Record = { en, zh, ja };\n\nexport function getTranslations(locale: string, namespace: string) {\n const messages = messagesMap[locale] || en;\n const ns = (messages as Record>)[namespace];\n const fallbackNs = (en as Record>)[namespace];\n return (key: string): string => {\n return ns?.[key] || fallbackNs?.[key] || key;\n };\n}\n"
},
{
"path": "web/src/lib/i18n.tsx",
"content": "\"use client\";\nimport { createContext, useContext, ReactNode } from \"react\";\nimport en from \"@/i18n/messages/en.json\";\nimport zh from \"@/i18n/messages/zh.json\";\nimport ja from \"@/i18n/messages/ja.json\";\n\ntype Messages = typeof en;\n\nconst messagesMap: Record = { en, zh, ja };\n\nconst I18nContext = createContext<{ locale: string; messages: Messages }>({\n locale: \"en\",\n messages: en,\n});\n\nexport function I18nProvider({ locale, children }: { locale: string; children: ReactNode }) {\n const messages = messagesMap[locale] || en;\n return (\n \n {children}\n \n );\n}\n\nexport function useTranslations(namespace?: string) {\n const { messages } = useContext(I18nContext);\n return (key: string) => {\n const ns = namespace ? (messages as any)[namespace] : messages;\n if (!ns) return key;\n return (ns as any)[key] || key;\n };\n}\n\nexport function useLocale() {\n return useContext(I18nContext).locale;\n}\n"
},
{
"path": "web/src/lib/utils.ts",
"content": "export function cn(...classes: (string | undefined | null | false)[]) {\n return classes.filter(Boolean).join(\" \");\n}\n"
},
{
"path": "web/src/types/agent-data.ts",
"content": "export interface AgentVersion {\n id: string;\n filename: string;\n title: string;\n subtitle: string;\n loc: number;\n tools: string[];\n newTools: string[];\n coreAddition: string;\n keyInsight: string;\n classes: { name: string; startLine: number; endLine: number }[];\n functions: { name: string; signature: string; startLine: number }[];\n layer: \"tools\" | \"planning\" | \"memory\" | \"concurrency\" | \"collaboration\";\n source: string;\n}\n\nexport interface VersionDiff {\n from: string;\n to: string;\n newClasses: string[];\n newFunctions: string[];\n newTools: string[];\n locDelta: number;\n}\n\nexport interface DocContent {\n version: string;\n locale: \"en\" | \"zh\" | \"ja\";\n title: string;\n content: string; // raw markdown\n}\n\nexport interface VersionIndex {\n versions: AgentVersion[];\n diffs: VersionDiff[];\n}\n\nexport type SimStepType =\n | \"user_message\"\n | \"assistant_text\"\n | \"tool_call\"\n | \"tool_result\"\n | \"system_event\";\n\nexport interface SimStep {\n type: SimStepType;\n content: string;\n annotation: string;\n toolName?: string;\n toolInput?: string;\n}\n\nexport interface Scenario {\n version: string;\n title: string;\n description: string;\n steps: SimStep[];\n}\n\nexport interface FlowNode {\n id: string;\n label: string;\n type: \"start\" | \"process\" | \"decision\" | \"subprocess\" | \"end\";\n x: number;\n y: number;\n}\n\nexport interface FlowEdge {\n from: string;\n to: string;\n label?: string;\n}\n"
},
{
"path": "web/tsconfig.json",
"content": "{\n \"compilerOptions\": {\n \"target\": \"ES2018\",\n \"lib\": [\"dom\", \"dom.iterable\", \"esnext\"],\n \"allowJs\": true,\n \"skipLibCheck\": true,\n \"strict\": true,\n \"noEmit\": true,\n \"esModuleInterop\": true,\n \"module\": \"esnext\",\n \"moduleResolution\": \"bundler\",\n \"resolveJsonModule\": true,\n \"isolatedModules\": true,\n \"jsx\": \"react-jsx\",\n \"incremental\": true,\n \"plugins\": [\n {\n \"name\": \"next\"\n }\n ],\n \"paths\": {\n \"@/*\": [\"./src/*\"]\n }\n },\n \"include\": [\n \"next-env.d.ts\",\n \"**/*.ts\",\n \"**/*.tsx\",\n \".next/types/**/*.ts\",\n \".next/dev/types/**/*.ts\",\n \"**/*.mts\"\n ],\n \"exclude\": [\"node_modules\"]\n}\n"
},
{
"path": "web/vercel.json",
"content": "{\n \"redirects\": [\n {\n \"source\": \"/:path(.*)\",\n \"has\": [\n {\n \"type\": \"host\",\n \"value\": \"learn-claude-agents.vercel.app\"\n }\n ],\n \"destination\": \"https://learn.shareai.run/:path\",\n \"permanent\": true\n },\n {\n \"source\": \"/\",\n \"destination\": \"/en\",\n \"permanent\": false\n }\n ]\n}\n"
}
]