agentleFS
Sign inStart free
AcKeskinAcKeskin/contexture/AGENTS.md

contexture

AGENTS.md2 starsChanged 6 days agoMITView on GitHub

Generated by skills/project-instructions/project-instructions.mjs from the discipline corpus. Do not hand-edit — edit architectural-rules/ and re-run. Vendor-neutral; read by Codex, Cursor, Copilot, and local models. (Claude Code uses CLAUDE.md + the corpus directly.) Be objective, direct, honest — not supportive-by-default, not reflexively contrarian. Weigh trade-offs and give a reasoned position; never just accept or reject. Surface weak reasoning, hidden assumptions, and the real cost of a choice even when unasked. A helper is most useful when honest, not agreeable. ✅ Always:…

  • Reads credentials

The file

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

4 rules. About 0 are common, 2 are distinctive, and 2 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.

  • 2

    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:

    Verify install / drift: node bootstrap/bootstrap.js --verify

    line 13
  2. Distinctive:

    Run the repo's test suite before committing.

    line 18
  3. Very specific:

    Read .claude/codemap.md for the file-by-file map (purpose + exports).

    line 22
  4. Distinctive:

    Never add AI attribution to commits or PRs (no Co-Authored-By, no "Generated with", no robot emoji).

    line 33

≈ “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

AGENTS.md

Generated by skills/project-instructions/project-instructions.mjs from the discipline corpus. Do not hand-edit — edit architectural-rules/ and re-run. Vendor-neutral; read by Codex, Cursor, Copilot, and local models. (Claude Code uses CLAUDE.md + the corpus directly.)

Stance (how to engage)

Be objective, direct, honest — not supportive-by-default, not reflexively contrarian. Weigh trade-offs and give a reasoned position; never just accept or reject. Surface weak reasoning, hidden assumptions, and the real cost of a choice even when unasked. A helper is most useful when honest, not agreeable.

Commands

  • Codemap (read before exploring): node skills/update-codemap/codemap.mjs writes .claude/codemap.md
  • Codemap diagrams: node skills/visualise-codemap/visualise-codemap.mjs
  • Verify install / drift: node bootstrap/bootstrap.js --verify
  • (Project-specific test/lint/build commands: see the repo README or ask — fill these in per repo.)

Testing

  • Run the repo's test suite before committing. Each change should be verifiable; a change with no way to tell if it worked is incomplete.

Project structure

  • Read .claude/codemap.md for the file-by-file map (purpose + exports). Read architectural-rules/ for the conventions. Read .claude/architecture.md if present.
  • Reusable skills are auto-discovered as Agent Skills from .claude/skills/ (run node bootstrap/bootstrap.js once after cloning to generate it; committed source is repo-root skills/).

Code style

  • Architectural rules live in architectural-rules/<scope>/: universal/ always applies; <language>/ applies when editing that language; config-authoring/ when editing skills/agents/rules/hooks. Read the ones matching what you touch.
  • Match existing style. Surgical edits — do not "improve" adjacent code.

Git workflow

  • One concern per commit; clear messages. Branch for feature work (feature/<slug>), merge --no-ff.
  • Never add AI attribution to commits or PRs (no Co-Authored-By, no "Generated with", no robot emoji).

Boundaries (the ✅ / ⚠️ / 🚫 block — projected verbatim)

✅ Always: read the codemap + applicable rules before coding; plan non-trivial work; one concern per commit; match existing style. ⚠️ Ask first: present a plan and wait for confirmation before non-trivial changes; before deleting pre-existing code you didn't write; before cross-cutting refactors. 🚫 Never: commit secrets / credentials / .env files; add AI attribution to commits or PRs (no Co-Authored-By, no "Generated with", no robot emoji); hardcode machine-specific paths or the owner's identity into shipped artefacts; force-push shared branches.

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