codybromcodybrom/clairvoyance/AGENTS.md

clairvoyance

AGENTS.md12 starsChanged 4 months agoMITView on GitHub

Guidance for AI coding agents working in this repository. For what Clairvoyance is and how to install it, see README.md. Run before committing: For site changes:

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 repository16 other files it gives its agents.
- abstraction-qualitySkill
- code-evolutionSkill
- comments-docsSkill
- complexity-recognitionSkill
- deep-modulesSkill
- design-it-twiceSkill
- design-reviewSkill
- diagnoseSkill
- error-designSkill
- general-vs-specialSkill
- information-hidingSkill
- module-boundariesSkill
- naming-obviousnessSkill
- pull-complexity-downSkill
- red-flagsSkill
- strategic-mindsetSkill

The file

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

3 rules. About 0 are common, 0 are distinctive, and 3 are 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.
3
Very specific
Tied to their project's names, files or tools. Probably skip.

Common
Distinctive
Very specific

Very specific:
Follow the writing-skills process: write a failing scenario first, then the skill, then verify.
line 18
Very specific:
Keep SKILL.md focused; move long reference material into references/.
line 20
Very specific:
Commit messages use conventional prefixes: feat:, fix:, docs:, chore:, evals:.
line 41

≈ “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

Guidance for AI coding agents working in this repository. For what Clairvoyance is and how to install it, see README.md.

Layout

skills/<name>/SKILL.md: the skills themselves, each with YAML frontmatter (name, description) and optional references/ files. This is the product.
agents/: subagent definitions used by skills (e.g. clean-room-alternative.md for design-it-twice).
evals/<skill>/: claude plugin eval suites. Each suite's README explains how to run it and what its graders score.
Platform manifests: .claude-plugin/, .codex-plugin/, .cursor-plugin/, .kimi-plugin/, .agents/, .codex/, .gemini/, .opencode/, gemini-extension.json, plugin.json, package.json.
docs/: the clairvoyance.fyi site (Astro). It reads skills/ at build time, so skill edits show up there automatically.
server.json: the hosted MCP server's entry for the official MCP Registry. Its version is kept in sync by scripts/bump-version.sh.
docs/edge/: the Cloudflare Worker in front of the site. It serves Markdown versions of pages via Accept: text/markdown and the read-only MCP server at /mcp. Cloudflare Workers Builds deploys it on every push to main. To deploy by hand, run npx wrangler deploy from that directory with CLOUDFLARE_ACCOUNT_ID set.

Writing skills

Skills translate A Philosophy of Software Design into checks an agent can run on real code. New or changed guidance should reinforce that philosophy, not general code-review advice.
Follow the writing-skills process: write a failing scenario first, then the skill, then verify.
A skill's description decides when it triggers. Changes to a description can change triggering for neighboring skills, so re-run the relevant eval suite.
Keep SKILL.md focused; move long reference material into references/.
The pillar each skill belongs to is defined in docs/src/utils/skills.ts. Add new skills there too.

Checks

Run before committing:

scripts/check-skills.sh          # skill and agent frontmatter consistency
scripts/bump-version.sh --check  # every manifest carries the same version
claude plugin validate . --strict

For site changes:

cd docs && npm ci && npm run build && npm test

Conventions

Commit messages use conventional prefixes: feat:, fix:, docs:, chore:, evals:.
Bump versions only with scripts/bump-version.sh <version>, never by hand, and record the change in CHANGELOG.md.
Eval runs cost real money (see each suite's README). Don't start one unless asked.

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