26zl26zl/universal-agent-skills/AGENTS.md

universal-agent-skills

AGENTS.md2 starsChanged 36 days agoMITView on GitHub

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 31,843 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 repository43 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- .github/copilot-instructions.mdCopilot instructions
- accessibility-auditSkill
- ai-feature-reviewSkill
- api-design-reviewSkill
- architecture-reviewSkill
- cli-and-scripts-auditSkill
- coding-styleSkill
- data-auditSkill
- data-minimizationSkill
- debugSkill
- describe-changesSkill
- destructive-ops-approvalSkill
- docs-consolidationSkill
- explain-codebaseSkill
- full-project-auditSkill
- implement-featureSkill
- improve-promptSkill
- incident-responseSkill
- infrastructure-auditSkill
- legal-compliance-reviewSkill
- license-complianceSkill
- no-ai-tracesSkill
- operations-auditSkill
- performance-auditSkill
- prepare-releaseSkill
- production-readiness-reviewSkill
- refactorSkill
- repository-auditSkill
- review-codeSkill
- secret-hygieneSkill
- security-auditSkill
- simplify-codeSkill
- start-projectSkill
- supply-chain-pinningSkill
- surgical-implementationSkill
- threat-modelSkill
- upgrade-dependenciesSkill
- ux-reviewSkill
- verify-changesSkill
- write-agent-instructionsSkill
- write-docsSkill
- write-testsSkill

The file

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

10 rules. About 0 are common, 4 are distinctive, and 6 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.
6
Very specific
Tied to their project's names, files or tools. Probably skip.

Common
Distinctive
Very specific

Very specific:
Keep canonical skills under skills/<name>/SKILL.md.
line 3
Very specific:
Keep installer behavior idempotent and refuse unmanaged conflicts unless --force is explicit.
line 4
Distinctive:
Prefer self-explanatory code.
line 5
Distinctive:
Keep comments factual, neutral, and normally one sentence.
line 6
Distinctive:
Keep external stack reconciliation dry-run by default, preserve unlisted plugins, and require explicit opt-in for persistent session …
line 7
Very specific:
Keep the hooks/guard.py rules high-precision.
line 8
Very specific:
Keep guard rules in hooks/guard.py only.
line 9
Very specific:
Keep the guard's shell wrapper as guard || exit 0.
line 11
Very specific:
Run python3 scripts/validate.py, python3 scripts/test_sync_agent_stack.py, python3 scripts/test_sync_instructions.py, python3 …
line 12
Distinctive:
Update both POSIX and PowerShell implementations when changing installer behavior.
line 13

≈ “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

Repository guidance

Keep canonical skills under skills/<name>/SKILL.md; do not maintain divergent agent-specific copies.
Keep installer behavior idempotent and refuse unmanaged conflicts unless --force is explicit.
Prefer self-explanatory code. Add comments only for non-obvious intent, invariants, constraints, workarounds, security boundaries, or surprising tradeoffs.
Keep comments factual, neutral, and normally one sentence. Never narrate prompts, AI use, the user-agent collaboration, debugging history, or obvious code behavior.
Keep external stack reconciliation dry-run by default, preserve unlisted plugins, and require explicit opt-in for persistent session capture.
Keep the hooks/guard.py rules high-precision; a false positive that blocks ordinary work gets the whole hook disabled.
Keep guard rules in hooks/guard.py only; hooks/opencode-guard.js maps OpenCode tool names onto the same payload and must not reimplement a rule.
Quote a guard violation in docs or tests with a uas-allow comment on its own line directly above it, never by splitting the literal into concatenated fragments.
Keep the guard's shell wrapper as guard || exit 0; interpreter script; the && … || exit 0 form turns the blocking exit code 2 into 0 and silently disables every rule.
Run python3 scripts/validate.py, python3 scripts/test_sync_agent_stack.py, python3 scripts/test_sync_instructions.py, python3 scripts/test_sync_opencode_config.py, python3 scripts/test_check_pin_freshness.py, python3 scripts/test_hooks.py, and the installer tests after relevant changes.
Update both POSIX and PowerShell implementations when changing installer behavior.

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