{"README.md": "Front door: what the kit is, the two editions, where to start", "HOW_TO_USE.md": "Operator manual: entry commands, the 0–6 lifecycle, how a run actually flows (NotebookLM canon source)", "USAGE.md": "Long-form usage narrative: worked patterns, wave discipline, watch instruments", "PRINCIPLES.md": "Narrative companion to kernel/principles.yaml (P1–P12)", "LESSONS.md": "Narrative companion to the scar file — human-readable stories behind the L-ids", "ORCHESTRATION.md": "Narrative companion to kernel/orchestration.yaml: phases, waves, monitors", "INTAKE.md": "Narrative companion to kernel/intake.yaml: the blocking-question doctrine", "CONTEXT_THRIFT.md": "Narrative companion to context-thrift: the conductor's token diet", "NOTEBOOKLM_SOURCE.md": "THE canon source for the NotebookLM visual suite (deck/infographic/mind-map)", "NOTEBOOKLM_PROMPTS.md": "The exact NotebookLM prompts + verbatim-terms QA contract for the visual suite", "feature-map.md": "Kit capability map — what exists, where it lives (NotebookLM source #2)", "FEATURE_MAP_BUILD_SPEC.md": "Spec that generated the feature map; regeneration contract", "kernel/principles.yaml": "P1–P12 — the doctrine. Lower id wins on conflict. CANONICAL", "kernel/lessons.yaml": "THE SCAR FILE — every lesson (incident + rules + ties), the kit's memory. CANONICAL, append-only, single-writer.", "kernel/domains.yaml": "Per-domain blind-spots/craft-moves (app, game, agent_tool, spec_packet, writing, data, ops); app domain carries the L114 auth deploy-gate", "kernel/executors.yaml": "Executor profiles (codex/claude/grok/agy/ollama/local) + the L115 escalation ladder + luna/terra/sol tiers + routing + L116 turnkey postures", "kernel/orchestration.yaml": "Fleet mechanics: phases 0–6, waves, control bus, monitors, side-delegation", "kernel/intake.yaml": "Intake question bank + conduct (spec-stage batch vs idea-stage dialogue)", "kernel/knowledge-base.yaml": "KB- atoms: durable how-to knowledge distilled from runs", "kernel/context-thrift.yaml": "CT rules — conductor context is the scarcest resource", "kernel/guardrail-admin.yaml": "How guardrails are authored/enforced as build-failing checks", "kernel/content-policy.yaml": "Content lanes incl. authorized_security_and_persona_register (L112 routing)", "kernel/ai-spine.yaml": "Governance spine: layers L0–L4, deny-by-default chain, BC dispositions", "kernel/machines.yaml": "Machine registry: mac main, macB (ssh extra), pi — reach + roles", "kernel/small-window.yaml": "Small-window operating mode (tight-context discipline; pinned in entabeni)", "skills/begin/SKILL.md": "/begin — boot a fleet: intake→apparatus (incl. L116 pool-spreading preflight)→pilot→fan-out→audit", "skills/continue/SKILL.md": "/continue — empirical re-entry: disk over memory, scoped tokens (L110)", "skills/handoff/SKILL.md": "[sensitive content omitted]", "skills/learn/SKILL.md": "/learn — the standing observer/harvester (this session's role)", "skills/dogfood/SKILL.md": "/dogfood — produce the human's manual walk plan after a completion", "skills/dogfood-it/SKILL.md": "/dogfood-it — the agent WALKS the product itself, evidence-backed", "skills/small-window/SKILL.md": "/small-window — enter/exit the tight-context mode", "skills/skin-it/SKILL.md": "/skin-it — motion-forward design-skin advisor (safe/bold/wildcard + :root tokens)", "skills/skin-it/skin-library.yaml": "20 aesthetic families + 16 motion patterns + 21 surface patterns + wildcard moves", "skills/skin-it/references/lovable-gallery.md": "~130 Lovable templates — the family grounding corpus", "skills/skin-it/references/envato-and-motion.md": "Envato UI kits + 2026 animated-web canon — the motion grounding", "skills/skin-it/references/surface-patterns.md": "Login/dashboard/etc archetypes — incl. Wolf's real Dribbble shots", "tools/dispatch.sh": "THE one multi-vendor launcher — per-vendor headless postures (the L116 fix lives here)", "tools/pool.sh": "Worker pool: launch/relaunch waves, per-slug executor overrides, no-progress breaker", "tools/assess.sh": "Classify every unit from ground truth (git+ledger+bus+process+marker)", "tools/audit-done.sh": "The DONE cross-check — no marker stands unaudited (L1)", "tools/dogfood-verify.sh": "Post-audit product walk verification", "tools/gate.sh": "Run a unit's build-failing gate", "tools/inject.sh": "Write a directive onto a worker's control bus", "tools/bus-sweep.sh": "Sweep control buses for unacked directives", "tools/keep-fed.sh": "Re-launch bounded-chunk executors (agy) until backlog drains", "tools/dashboard.sh": "Live per-agent terminal view", "tools/progress.sh": "Fleet % from ground truth", "tools/watch.sh": "Arm exception/heartbeat/progress monitors", "tools/lib.sh": "Shared shell helpers for the tool set", "tools/normalize-report.sh": "Unwrap/validate vendor report envelopes (grok json etc.)", "tools/verdict.sh": "NotebookLM verdict lane helper (claim-vs-canon)", "tools/offload.sh": "Move heavy artifacts out of the conductor's context path", "tools/claude-all-watch.py": "Observer watcher: every local Claude session → HUMAN/OUTCOME/APIERROR lines", "tools/codex-all-watch.py": "Observer watcher: every codex rollout (interactive + exec fleets)", "tools/observify-inbox-watch.py": "Observify interjection consumer — a session tails its own inbox (+ack)", "templates/worker-prompt.md": "The worker prompt skeleton with the INTEGRITY clause", "templates/project.env": "Per-project fleet config consumed by the tools", "templates/CONTROL.yaml": "Control-bus file format (directives, acks)", "templates/STATUS.md": "Per-unit status ledger skeleton", "templates/handoff-packet.md": "The dispatch packet — no packet, no dispatch", "templates/guardrails.yaml": "Guardrail declaration skeleton (build-failing)", "templates/pipeline.yaml": "Multi-stage unit pipeline skeleton", "templates/report.schema.json": "Schema-forced vendor report shape", "templates/ctx-check.sh": "Context-threshold self-check (L94 gate)", "templates/statusline-context.sh": "Statusline context% surfacing", "templates/sandboxed-claude.sh": "Sandboxed claude worker launcher", "templates/worker-sandbox.sb": "macOS seatbelt profile for sandboxed workers", "templates/authorized-context-header.md": "L112 true-context header (security engagement / authored persona)", "examples/fable-terminal-build/project.env": "Worked example: a real project.env", "examples/fable-terminal-build/worker-prompt.md": "Worked example: a real instantiated worker prompt", "build-fable-terminal/README.md": "Sub-kit: fable-terminal build program (spec/backlog/eval/tools)", "build-fable-terminal/SPEC.yaml": "fable-terminal: canonical product spec", "build-fable-terminal/BACKLOG.yaml": "fable-terminal: ticket backlog", "build-fable-terminal/EVAL.yaml": "fable-terminal: eval harness spec", "build-fable-terminal/TOOLS.yaml": "fable-terminal: tool surface spec", "build-fable-terminal/ARCHITECTURE.md": "fable-terminal: architecture doc", "pi-fable-terminal/README.md": "Sub-kit: the Pi deployment of fable-terminal", "pi-fable-terminal/PI_SPEC.yaml": "Pi: deployment spec", "pi-fable-terminal/SETUP.yaml": "Pi: setup steps", "pi-fable-terminal/RASPBIAN_KNOWLEDGE.yaml": "Pi: hard-won raspbian facts", "pi-fable-terminal/FINISH_RUNBOOK.md": "Pi: the finish runbook (unblocked 07-08)", "skills/pi/SKILL.md": "GOD-ONLY /pi — the hardware lane: Pi deploy/finish, raspbian fact bank, on-device proof", "skills/machines/SKILL.md": "GOD-ONLY /machines — registry reach/health/roles; resolve, never fish (L93)", "skills/ship/SKILL.md": "GOD-ONLY /ship — L34-guarded fleet ship: mirrors + both zips + live deploy, verified", "skills/harvest/SKILL.md": "GOD-ONLY /harvest — write one scar right: id, incident, rules, ties, encode, ship", "skills/verdict/SKILL.md": "GOD-ONLY /verdict — claim-vs-canon via the NotebookLM lane, never self-adjudicated", "skills/guardrails/SKILL.md": "GOD-ONLY /guardrails — build-failing guards: name, class, prove RED, install", "skills/drift/SKILL.md": "GOD-ONLY /drift — scan + classify divergence: resync vs harvest-first vs promote"}