quiver
AGENTS.md16 starsChanged 57 days agoMITView on GitHub
Read CONTRIBUTING.md before changing anything. CLAUDE.md has the architecture and the per-component rules.
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 31,439 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 repository23 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- adviseSkill
- brainstormSkill
- code-navigationSkill
- commitSkill
- create-agents-mdSkill
- create-prSkill
- design-buildSkill
- design-fixSkill
- designSkill
- handoverSkill
- hypothesis-debuggingSkill
- load-handoverSkill
- local-filesSkill
- planSkill
- report-checkSkill
- reviewSkill
- senior-reviewSkill
- shipSkill
- tddSkill
- verificationSkill
- visual-companionSkill
- workSkill
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 3
Read CONTRIBUTING.md before changing anything.
≈ “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
Read CONTRIBUTING.md before changing anything. CLAUDE.md has the architecture and the per-component rules.
Must-follow constraints
- Branching: Create a feature branch for any non-trivial change. If already on the correct branch, keep using it -- do not create additional branches or worktrees unless explicitly requested.
- Safety: Do not delete or overwrite user data. Avoid destructive commands.
- Tests: Run
bash tests/run-all.shbefore you finish. Most tests are contract checks between two copies of one string; a red test points at the copy you have not updated. Fix the copy, not the test. - ASCII-first: Use ASCII characters only unless the file already contains Unicode.
- Release: Plugin version lives in
.claude-plugin/plugin.json"version"field. Release is handled by a private project command -- do not bump version, create tags, or publish releases manually.
Important locations
.claude-plugin/plugin.json-- Plugin manifest (name, version, skill/agent registration)..claude-plugin/marketplace.json-- Marketplace listing metadata (keywords, tags, category).hooks/hooks.json-- Hook event registration: PreToolUse (pre-tool-use-guard.sh, classifies a Bash command as deny/ask/silence), PreCompact (pre-compact-handover.sh, saves a handover), SessionStart (session-start-auto-dispatch.sh, emits the skill routing block).hooks/scripts/pre-compact-handover.sh-- PreCompact hook implementation; usesclaude -pfor summarization.skills/handover/SKILL.md-- the 8 handover headings sit under aSYNC:comment that points at the hook script.hooks/scripts/pre-compact-handover.sh-- the same 8 headings insidePROMPT_PREFIX, under aSYNC:comment that points back.tests/hooks/test-handover-sync-contract.shfails when the two lists differ.skills/local-files/SKILL.md-- the rule that keeps.claude/plans/,.claude/reports/,.claude/specs/,.claude/handovers/,.claude/work/, and.claude/ship/out of git through.git/info/exclude.hooks/scripts/pre-compact-handover.shcarries its one bash copy.tests/skills/test-local-files-contract.shchecks that each writer on its list names the rule;tests/hooks/test-handover-local-exclude.shruns the hook's copy.
Change safety rules
- Changing the timestamp format breaks lexicographic sort ordering of existing handover files.
Known gotchas
- The PreCompact hook pipes transcript + prompt to
claude -pvia stdin to avoidARG_MAXlimits -- do not refactor to use command-line arguments. - Hook timeout is 180 seconds (
hooks.json), but the innerclaude -pcall has atimeout 150-- the 30-second gap prevents zombie processes. ls -1rin the prune loop relies on filenames being timestamps for correct sort order -- non-timestamp filenames will break pruning.- The hook silently exits 0 on any failure (missing transcript, empty claude output) -- check handover directory contents to verify it actually ran.
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.

