agentleFS
Sign inStart free
maya8237maya8237/MEOW/AGENTS.md

MEOW

AGENTS.md3 starsChanged 34 days agoMITView on GitHub

Claude Agent SDK harness (meow) and Claude Code plugin (skills/). From the repo root with .venv active (uv pip install -e ".[dev]", or python -m pip install -e ".[dev]" without uv): Use pytest for both unittest and function tests. Root imports use the installed src/ package; linked worktrees need PYTHONPATH=src. Read only the relevant guide/section. Keep maintained docs/skills below 200 lines; remove repetition and link topic-specific details instead of loading everything. Reduce words too; do not meet the line limit…

  • Installs packages
  • Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.

    Of the AGENTS.md files we read, the 31,774 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.

  • In this repository9 other files it gives its agents.

The file

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

5 rules. About 0 are common, 2 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use pytest for both unittest and function tests.

    line 16
  2. Distinctive:

    Read only the relevant guide/section.

    line 21
  3. Very specific:

    Use the narrowest owning package.

    line 39
  4. Distinctive:

    Keep native/CLI guidance aligned.

    line 47
  5. Very specific:

    Use project/plan_files.py helpers plan_review_file, plan_test_file, new_review_filename.

    line 54

≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.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
AGENTS.md

MEOW

Claude Agent SDK harness (meow) and Claude Code plugin (skills/).

Development

From the repo root with .venv active (uv pip install -e ".[dev]", or python -m pip install -e ".[dev]" without uv):

python -m pytest -q
ruff check .
ruff format --check .

Use pytest for both unittest and function tests. Root imports use the installed src/ package; linked worktrees need PYTHONPATH=src.

Context and references

Read only the relevant guide/section. Keep maintained docs/skills below 200 lines; remove repetition and link topic-specific details instead of loading everything. Reduce words too; do not meet the line limit by minifying prose.

  • CLI: command index; meow COMMAND --help gives syntax.
  • Integrations: configuration and optional services.
  • Workflows: definitions, scopes, templates, recovery.
  • Tester setup: browser, dev server, inference, preflight.
  • Hooks: user hook definitions and meow native hook.
  • Architecture: package/module ownership.
  • Native protocol: in-session helpers.

Config priority: .meow/config.local.toml > .meow/config.toml > ~/.meow/config.toml. Runtime evidence/journals live in .meow/, plans default to .meow/plans/, and project docs stay in docs/.

Ownership

Use the narrowest owning package; package roots contain entry points. Role prose belongs in src/meow/project/custom/builtin/agents/*.toml; project/prompts.py owns computed context/verdict contracts.

skills/*/SKILL.md owns plugin flows. Run defaults to CLI; plan/review/lint default to native. Orchestrator tasks remain detached CLI sprints. Schedule uses OS/CLI; onboard adds optional setup, and customize edits definitions. Keep native/CLI guidance aligned.

Invariants

  • SDK allowed_tools bypasses can_use_tool; enforce with agents.base.guard_tools.
  • Verdicts start SUMMARY:, then STATUS: PASS|FAIL; parse only with agents.reviewer._verdict_status.
  • Use project/plan_files.py helpers plan_review_file, plan_test_file, new_review_filename; never reconstruct artifact names.
  • CLI/native orchestration share execution/orchestration/tick.py. Push only in delivering; task refs use orchestrator/<id>-<task>, never orchestrator/<id>/<task>, beside integration orchestrator/<id>.
  • run_sprint settles pre-terminal exceptions as parked unless blocking (execution/run_failures.py); inner steps record phases only.
  • Definitions replace whole files: local > project > user > builtin. Gates/permissions remain code-owned; user hooks may only deny or inform.
  • Claude owns transcripts; journals store role/session IDs, resumed only by meow resume.

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.

Check your setup, free