agentleFS
matt82198matt82198/aesop/CLAUDE.md

aesop

CLAUDE.md2 starsChanged 31 days agoMITView on GitHub

What: MIT-licensed orchestration kernel for autonomous software development (v0.8.0-v0.11.0 were released under PolyForm Noncommercial 1.0.0; v0.7.0 and earlier under MIT; MIT applies from the 2026-10-26 relicense forward). RED CI runs (any FAILURE check on main or open PRs) are detected whenever the orchestrator runs idle_tick (no scheduled cadence) by ~/scripts/detect_red_ci_runs.py, which idle_tick invokes. Detected escapes are queued to ~/.claude/gate-escapes-queue.json; idle_tick surfaces them as HIGH priority; orchestrator's main thread invokes gate-escape-resolver workflow to fix + postmortem + enforce. No RED run…

The file

We split this CLAUDE.md into the rules it gives an agent

1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    No RED run is missed; deduplication prevents redundant queue entries (keys on run_id).

    line 11

≈ “Common” is an estimate: at least 1 in 5 of the 27,641 CLAUDE.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
CLAUDE.md

Aesop Kernel — Project CLAUDE.md

What: MIT-licensed orchestration kernel for autonomous software development (v0.8.0-v0.11.0 were released under PolyForm Noncommercial 1.0.0; v0.7.0 and earlier under MIT; MIT applies from the 2026-10-26 relicense forward).

RED CI Escape Detection

RED CI runs (any FAILURE check on main or open PRs) are detected whenever the orchestrator runs idle_tick (no scheduled cadence) by ~/scripts/detect_red_ci_runs.py, which idle_tick invokes. Detected escapes are queued to ~/.claude/gate-escapes-queue.json; idle_tick surfaces them as HIGH priority; orchestrator's main thread invokes gate-escape-resolver workflow to fix + postmortem + enforce. No RED run is missed; deduplication prevents redundant queue entries (keys on run_id).

Domain map

  • skills/ — Orchestration skills (/power: priming skill) — read skills/CLAUDE.md
  • daemons/ — Watchdog daemon (repo backup, secret-scan gate, heartbeat) — read daemons/CLAUDE.md
  • dash/ — TUI dashboard (watchdog-gui.sh, real-time fleet status) — read dash/CLAUDE.md
  • monitor/ — Orchestration monitor (signals, AUTO/PROPOSE logic) — read monitor/CLAUDE.md
  • mcp/ — Read-only MCP server (fleet status, agents, tracker, costs) — read mcp/CLAUDE.md
  • scan/ — Example IOC/secret scanner template — read scan/CLAUDE.md
  • tools/ — Build utilities (Python/shell, secret scanning, verification gates) — read tools/CLAUDE.md
  • hooks/ — Git pre-push policy enforcement (branch protection, secrets) — read hooks/CLAUDE.md
  • driver/ — AgentDriver multi-model seam (Claude Code, Codex, OpenAI-compatible backends; wave bridge for verification routing) — read driver/CLAUDE.md
  • driver/orchestrator-swap/ — Orchestrator-backend selection, configuration, and seat-swap (HS-1/HS-2 increments; wave scheduler; adjudication gate) — read driver/orchestrator-swap/CLAUDE.md
  • bin/ — CLI scaffolder (Node.js entry point) — read bin/CLAUDE.md
  • stdio/ — Generic zero-dependency stdin/stdout pipeline (byte-exact filter, framing, stream-json, TTY seam) — read stdio/CLAUDE.md
  • chat/ — Zero-dependency interactive chat TUI (aesop chat; pure render and input reducer, streaming session, TTY seam via stdio/) — read chat/CLAUDE.md
  • auth/ — Generic zero-dependency auth and model selection (provider registry, single JSON credential store, env-beats-store precedence, auth/model verbs and control channel) — read auth/CLAUDE.md
  • ui/ — Web dashboard (SSE, CSRF protection, collector thread) — read ui/CLAUDE.md
  • state_store/ — Event-sourced state layer (SQLite WAL, projections) — read state_store/CLAUDE.md
  • bench/ — Held-out model benchmark (quality scorer) — read bench/README.md
  • templates/ — Wave-manifest presets (saas/data/library JSON) consumed by tools/wave_templates.py — read tools/CLAUDE.md
  • tests/ — Test suites (shell, Node, Python) and fixtures — read tests/CLAUDE.md
  • docs/ — Architecture guides, tutorials, setup — read docs/
  • assets/ — Logo, branding, media
  • state/ — Runtime durable checkpoints (git-ignored)
  • examples/ — Reference implementations and adopter walkthrough kits — read examples/first-wave-baseline/README.md
  • scripts/ — Repo maintenance wrappers (stats verification) — see § scripts/ below

examples/

Adopter reference implementations and verified baseline walkthroughs. Artifacts here demonstrate the aesop product end-to-end: canonical wave cycles, manifest schema compliance, and realistic onboarding flows. Invariants: (1) All manifests validate against tools/wave_templates.py schema; (2) No personal paths, secrets, or invented CLI flags; (3) All claimed commands must be verifiable from bin/cli.js --help or repo tools; (4) README files include honest timing, cost, and prerequisites.

scripts/

Thin CRLF-safe shell wrappers for repo maintenance. Currently: verify-stats.sh — the single-source stats gate (delegates to tools/self_stats.py --check / --regenerate; stats.json is the only stats source consumed by README and the portfolio). Invariants: wrappers delegate to tools/, never own logic; ASCII, no line continuations; exit codes propagate (fail-closed).

Dispatch rule

Workers read exactly ONE domain CLAUDE.md; this file is navigation only.

  • LANE-CONTRACT.md — the lane-contract hook refuses any dispatch whose prompt does not reference it (it injects nothing); read it before writing code.

Setup for development

See docs/ for full setup, architecture, and usage guides.

Relocated

  • Key principles (Subagents always Haiku, Orchestrator on main thread only, State committed to git, Secret-scan gates, Idempotent + append-only, Observable machinery) → inlined into each domain's CLAUDE.md
  • Branch + PR discipline (feature/* only, secret-scan.py gate, not a vault repo) → tools/CLAUDE.md
  • Setup for development (install, config, watchdog test, dashboard, monitor extension) → docs/INSTALL.md

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.