agents-lab
AGENTS.md0 starsChanged 3 months agoMITView on GitHub
After code or configuration changes:
- Reads credentials
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 repository19 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- analyzeSkill
- buildSkill
- data-pipelineSkill
- handoffSkill
- orchestrateSkill
- plan-taskSkill
- plugin-devSkill
- reviewSkill
- verifySkill
- analyzeSkill
- buildSkill
- data-pipelineSkill
- handoffSkill
- orchestrateSkill
- plan-taskSkill
- plugin-devSkill
- reviewSkill
- verifySkill
The file
We split this AGENTS.md into the rules it gives an agent
8 rules. About 0 are common, 5 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈5
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
- Distinctive: line 5
Read a file before editing it.
- Distinctive: line 8
Do not modify adjacent code or formatting without a task reason.
- Very specific: line 9
Never commit secrets, private data, logs, model files, .env, local state, or credentials.
- Distinctive: line 10
Keep implementation files under 500 lines.
- Very specific: line 16
Read FILESYSTEM.md before creating a new path or cross-agent artifact.
- Distinctive: line 25
Do not claim execution, tests, or validation without current evidence.
- Very specific: line 32
Never edit a generated mirror directly.
- Distinctive: line 36
After code or configuration changes:
≈ “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
Working principles
- Read a file before editing it.
- State assumptions and surface material ambiguity.
- Make the smallest change that satisfies the task contract.
- Do not modify adjacent code or formatting without a task reason.
- Never commit secrets, private data, logs, model files,
.env, local state, or credentials. - Keep implementation files under 500 lines.
Project state
TASKS.mdis the only task blackboard.REGISTRY.mdis the registry for agents, skills, plugins, commands, and durable outputs.- Read
FILESYSTEM.mdbefore creating a new path or cross-agent artifact. artifacts/contains Markdown messages only.
Separation of powers
- Creators implement; reviewers review read-only; verifiers validate evidence.
- Review findings use CRITICAL, HIGH, MEDIUM, or LOW severity.
- CRITICAL or HIGH findings block completion.
- A reviewer returns its report to the parent agent. The parent persists it unchanged under
artifacts/review/. - Do not claim execution, tests, or validation without current evidence.
Runtime sources
.codex/agents/is the canonical agent source..agents/skills/is the canonical skill source..claude/agents/,.claude/skills/, and plugin skills are generated mirrors.- Never edit a generated mirror directly; run
agents-lab sync-runtimes --write.
Verification
After code or configuration changes:
- Run focused tests.
- Run
agents-lab sync-runtimes --check. - Run
agents-lab validate --strict. - Obtain independent
code-reviewerreview. - Fix and re-review until there are no CRITICAL/HIGH findings.
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.