.agents
AGENTS.md0 starsChanged 6 days agoMITView on GitHub
This public repository holds Weaverse's team rules, first-party skills and optional Claude Code glue. The shared instructions live in rules/weaverse-team.md; follow them when changing this repository.
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 32,671 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 repository12 other files it gives its agents.
- agent-work-verificationSkill
- create-taskSkill
- database-migration-reviewSkill
- debug-logsSkill
- dependency-upgradesSkill
- fix-prSkill
- pr-merge-gatesSkill
- production-debuggingSkill
- release-verificationSkill
- reliability-reviewSkill
- shipSkill
- workSkill
The file
We split this AGENTS.md into the rules it gives an agent
10 rules. About 1 is common, 4 are distinctive, and 5 are very specific to this project.
≈1
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
5
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 7
Put it in skills/<lowercase-hyphen-name>/SKILL.md and add its folder to the .gitignore allowlist.
- Very specific: line 10
Keep metadata a string-to-string map.
- Very specific: line 12
Avoid built-in command names.
- Distinctive: line 13
Keep each skill under 150 lines where practical.
- Distinctive: line 16
Keep only team contracts, executable helpers, current platform traps or verification a model would miss.
- Common: line 19
Keep customer data, personal names, private paths, identifiers and credentials out of every public file.
- Very specific: line 23
Test installer changes only with a throwaway HOME.
- Very specific: line 24
Test notifications with --dry-run.
- Distinctive: line 25
Keep shell files compatible with Bash 3.2 and check them with ShellCheck.
- Distinctive: line 26
Update the README catalogue and migration notes with behavior changes.
≈ “Common” is an estimate: at least 1 in 5 of the 41,581 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
Contributing to .agents
This public repository holds Weaverse's team rules, first-party skills and optional Claude Code glue.
The shared instructions live in rules/weaverse-team.md; follow them when changing this repository.
Add a skill
- Put it in
skills/<lowercase-hyphen-name>/SKILL.mdand add its folder to the.gitignoreallowlist. - Match
nameto the folder (1 to 64 ASCII lowercase letters, digits and single hyphens). - Set a string
descriptionof at most 1024 characters, including a specific "Use when" trigger. - Keep
metadataa string-to-string map; useallowed-toolsfor tool hints, notmetadata.tools. - Set
disable-model-invocation: truefor workflows that only a person should initiate. - Avoid built-in command names;
scripts/claude-commands.jsonrecords the checked collision list. - Keep each skill under 150 lines where practical; 500 lines is the validation ceiling.
Apply the frontier filter
- Keep only team contracts, executable helpers, current platform traps or verification a model would miss.
- Cut generic advice and incident narratives. Explain the non-obvious failure, not textbook practice.
- Reference maintained upstream skills with pinned sources. Do not copy third-party skills here.
- Keep customer data, personal names, private paths, identifiers and credentials out of every public file.
Verify and maintain
- Install the pinned validation dependencies as described in
README.md, then runbash scripts/validate.sh. - Test installer changes only with a throwaway
HOME; use--dry-runbefore any real installation. - Test notifications with
--dry-run; never send a desktop notification from a test. - Keep shell files compatible with Bash 3.2 and check them with ShellCheck.
- Update the README catalogue and migration notes with behavior changes.
- Review quarterly; use upstream skill-creator evaluations before major skill changes or model/tool upgrades.
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.

