agentleFS
Sign inStart free
mojopolymojopoly/three-loops/templates/CLAUDE.md

three-loops / templates

CLAUDE.md0 starsChanged 3 months agoMITView on GitHub

{{One paragraph: what this is, the stack, where things live.}} Read CONTEXT.md before non-trivial work — it holds the domain glossary, invariants, and the shared-resource registry. Use its terms exactly. Read PRINCIPLES.md and honor it absolutely: its ask-first list overrides any instruction to proceed autonomously, and its product principles are review criteria. After any material change: launch, drive the changed flow end to end, watch the logs, exercise one failure path, and capture proof for the PR body. Generated code…

  • Reads credentials

The file

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

7 rules. About 0 are common, 4 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    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
  • could be automatic: a setting could enforce it
  1. Very specific:

    Read CONTEXT.md before non-trivial work — it holds the domain glossary, invariants, and the shared-resource registry.

    line 9
  2. Very specific:

    Read PRINCIPLES.md and honor it absolutely: its ask-first list overrides any instruction to proceed autonomously, and its product …

    line 11
  3. Distinctive:

    After any material change: launch, drive the changed flow end to end, watch the logs, exercise one failure path, and capture proof for the …

    line 21
  4. Distinctive:

    Commit bodies say why when the diff doesn't.

    line 30
  5. Distinctive:

    Never build on one without re-verifying it against the primary source.

    line 43
  6. Very specific:

    Never commit secrets; .env* files stay untracked.

    line 59
  7. Distinctive:

    BEFORE writing one; check open branches for competing claims.

    line 61

≈ “Common” is an estimate: at least 1 in 5 of the 26,980 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

CLAUDE.md — agent operating rules for this repository

<!-- Template: fill every {{placeholder}}, delete sections that don't apply. -->

Project

{{One paragraph: what this is, the stack, where things live.}}

Read CONTEXT.md before non-trivial work — it holds the domain glossary, invariants, and the shared-resource registry. Use its terms exactly. Read PRINCIPLES.md and honor it absolutely: its ask-first list overrides any instruction to proceed autonomously, and its product principles are review criteria.

Runtime

  • Launch: {{command — e.g. docker compose up -d --wait && pnpm dev}}
  • Logs: {{command or location}}
  • Tunnel (OAuth/webhooks): {{./dev-tunnel.sh — see templates/dev-tunnel.sh}}

After any material change: launch, drive the changed flow end to end, watch the logs, exercise one failure path, and capture proof for the PR body. Generated code that has never executed is a draft.

Commits

  • Small and intentional: feat: / fix: / test: / refactor: / docs:, each reviewable alone. Never one giant drop.
  • Tests land in the same PR as the behavior they pin.
  • Commit bodies say why when the diff doesn't.

Tests

  • Run: {{command}}
  • Unit tests are dependency-free: no network, no database, injectable clock.
  • New logic-heavy code (time math, state transitions, policy) is built test-first. Glue and layout are covered by live verification instead.

Tripwires

Comments prefixed VOLATILE: mark facts that drift independently of this repo (external API limits, provider model names, prices, framework quirks). Never build on one without re-verifying it against the primary source. When you fix a bug caused by such a fact, leave a new tripwire at the site.

Framework guards

<!-- Keep a guard block per framework whose current version postdates common knowledge. Example: --><!-- BEGIN:framework-guard -->

{{Framework}} {{version}} in this repo may differ from your training data — APIs, conventions, and file structure included. Before writing code against it, read the local docs: {{path, e.g. node_modules/<pkg>/docs/}}. Heed deprecation warnings over remembered idioms.

<!-- END:framework-guard -->

Boundaries

  • Never commit secrets; .env* files stay untracked.
  • Migrations: claim your slot in CONTEXT.md § Shared-resource registry BEFORE writing one; check open branches for competing claims.
  • Destructive operations (drops, deletes, force-pushes): propose, don't run.

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.

Reviews can't be read right now.

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