skills / templates
AGENTS.md9 starsChanged 31 days agoMITView on GitHub
Instructions for AI agents (Claude Code, Codex, Cursor and others) and a fast orientation for humans. CLAUDE.md should contain one line, @AGENTS.md, so every tool reads the same file. Read WORKING_AGREEMENTS.md. The short version: - The base branch is main. Branch off it, PR back into it. - Agents may push branches and open draft PRs. Agents never merge, deploy, or push to main. A human pulls the trigger. Approval in one message does not carry to the next. -…
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 31,542 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 repository13 other files it gives its agents.
- announcing-a-featureSkill
- authoring-a-prdSkill
- building-componentsSkill
- design-tokensSkill
- docs-sync-reviewSkill
- figma-buildSkill
- figma-importSkill
- landing-a-design-system-changeSkill
- mobile-responsive-reviewSkill
- pre-push-reviewSkill
- starting-a-changeSkill
- testing-a-changeSkill
- updating-the-changelogSkill
The file
We split this AGENTS.md into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 12
Read WORKING_AGREEMENTS.md.
≈ “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.
Files
AGENTS.md
Instructions for AI agents (Claude Code, Codex, Cursor and others) and a fast orientation for
humans. CLAUDE.md should contain one line, @AGENTS.md, so every tool reads the same file.
What this repo is
<!-- One paragraph. What it is, what it is NOT (no production data? not published?), who reviews. -->The rules that don't bend
Read WORKING_AGREEMENTS.md. The short version:
- The base branch is
main. Branch off it, PR back into it. - Agents may push branches and open draft PRs. Agents never merge, deploy, or push to
main. A human pulls the trigger. Approval in one message does not carry to the next. - One concern per PR. No stacked PRs.
- Green
npm run preflightbefore opening a PR. - Secrets come from the environment, never a literal in source.
Start and finish a change (use the skills)
- Writing a spec:
authoring-a-prd. - Before branching:
starting-a-change(branch naming, fresh base, latestRETRO.mdtraps). - Building UI:
figma-import→building-components→figma-build. Tokens:design-tokens. - While building and before pushing:
testing-a-change(unit, integration and component tests at the right layer, with independent expected values). - Before pushing:
pre-push-review(self-review againstdocs/standards/STANDARDS.md, then the local gates). It callsupdating-the-changelog,announcing-a-feature,docs-sync-reviewandmobile-responsive-reviewwhere they apply. - After a design-system merge:
landing-a-design-system-change. A merge is not delivery.
Configuration the skills read
| Key | Value |
| Base branch | main |
| Branch areas | feat fix ui docs ci chore |
| Preflight | npm run preflight |
| Unit tests | npm test |
| Integration tests | npm run test:integration (needs npm run dev) |
| Component a11y tests | npm run test:storybook |
| Changelog | CHANGELOG.md (or a folder, one file per entry) |
| Feature announcements | <!-- e.g. lib/feature-announcements/entries.ts, 45-day window --> |
| In-app docs map | docs/docs-map.md |
| Product truth for PRDs | <!-- e.g. docs/prd/_features.md --> |
| Exemplar component | <!-- e.g. src/ui/button.tsx --> |
Where things live
<!-- app/, components/, lib/, docs/decisions/, docs/standards/ … -->Known traps
See the newest entries in RETRO.md.
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
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.