dotcursor
AGENTS.md6 starsChanged 27 days agoMITView on GitHub
Reusable Agent Skills and always-on rules for PRs, review, and shipping. Skills install through the Agent Skills CLI. Rule bodies live in rules/*.mdc. Claude Code reads this file when CLAUDE.md is absent. Use it for install and conventions. This file does not apply the .mdc rules. Do not paste rule bodies here. Skills (any Agent Skills host): npx skills add does not install rules. Copy rules/*.mdc if you want the always-on kit loop. How each vendor picks up rules is…
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 33,315 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 repository20 other files it gives its agents.
- rules/AGENTS.mdAGENTS.md
- kit-audit-metricsSkill
- kit-audit-skillsSkill
- kit-capture-lessonSkill
- kit-challenge-designSkill
- kit-check-blastSkill
- kit-create-prSkill
- kit-debate-decision-treeSkill
- kit-debug-prodSkill
- kit-decisionsSkill
- kit-evolveSkill
- kit-explainSkill
- kit-learn-repoSkill
- kit-minimizeSkill
- kit-optimize-contextSkill
- kit-planSkill
- kit-proveSkill
- kit-reviewSkill
- kit-strip-diffSkill
- kit-triage-commentsSkill
The file
Show its rules and text
We split this AGENTS.md into the rules it gives an agent
21 rules. About 1 is common, 9 are distinctive, and 11 are very specific to this project.
≈1
Common · estimate
Many files like this have a rule of this kind.
≈9
Distinctive · estimate
This author's own choices. Worth a look.
11
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
- Very specific: line 26
Always-on rules are not slash skills.
- Distinctive: line 45
When adding a new skill:
- Very specific: line 46
Create a directory under skills/ with the kit- prefix (e.g., skills/kit-new-skill/)
- Very specific: line 47
Add a SKILL.md file with YAML frontmatter:
- Distinctive: line 55
Update the Skills Catalog table in README.md
- Very specific: line 56
Run kit-audit-skills (overlaps, stale kit-* refs, catalog vs disk, npx skills add .
- Distinctive: line 59
When removing a skill:
- Very specific: line 60
Delete the skill directory from skills/
- Distinctive: line 61
Remove the entry from the Skills Catalog table in README.md
- Distinctive: line 64
When renaming a skill:
- Very specific: line 66
Update the name field in the YAML frontmatter
- Very specific: line 74
Create rules/kit-<name>.mdc with YAML frontmatter (description, alwaysApply and/or globs) and the rule body
- Distinctive: line 75
Keep each rule to one concern and under ~50 lines
- Very specific: line 76
Add a row to rules/AGENTS.md
- Distinctive: line 77
Update the Rules Catalog in README.md
- Very specific: line 80
Delete or rename the .mdc file
- Very specific: line 81
Update rules/AGENTS.md and the Rules Catalog in README.md
- Common: line 85
No secrets: Never commit credentials, tokens, private hostnames, or work emails
- Distinctive: line 86
Keep skills generic: Avoid employer names, internal tools, and proprietary details.
- Distinctive: line 87
Maintain catalog: Always update README.md when changing skills or rules
- Very specific: line 88
Installation path: Use netanelavr/agent-kit in all documentation examples
≈ “Common” is an estimate: at least 1 in 5 of the 42,283 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
agent-kit
Reusable Agent Skills and always-on rules for PRs, review, and shipping. Skills install through the Agent Skills CLI. Rule bodies live in rules/*.mdc.
Claude Code reads this file when CLAUDE.md is absent. Use it for install and conventions. This file does not apply the .mdc rules. Do not paste rule bodies here.
Install
Skills (any Agent Skills host):
npx skills add netanelavr/agent-kit --allOne skill:
npx skills add netanelavr/agent-kit/skills/kit-create-prnpx skills add does not install rules. Copy rules/*.mdc if you want the always-on kit loop. How each vendor picks up rules is in the README consumption matrix.
Use
- Invoke a skill with
/<skill-name>(Cursor) or the host’s skill command. - Always-on rules are not slash skills. Bodies:
rules/*.mdc. How each vendor attaches them: README section How vendors consume the kit.
Kit conventions
- Skill directories and names use the
kit-prefix (skills/kit-<name>/). - Rule ids use the same prefix (
kit-comment-why, …). Bodies live inrules/kit-<name>.mdc. - Skills must stay generic: no employer names, internal tools, or proprietary details.
- Public examples use
netanelavr/agent-kit. - Catalogs in
README.mdmust match what is on disk.
Agent Maintenance Guide
Guidelines for maintaining agent-kit.
Skill Management
Adding New Skills
When adding a new skill:
- Create a directory under
skills/with thekit-prefix (e.g.,skills/kit-new-skill/) - Add a
SKILL.mdfile with YAML frontmatter:--- name: kit-new-skill description: Clear description of when to use this skill disable-model-invocation: true --- - Update the Skills Catalog table in
README.md - Run
kit-audit-skills(overlaps, stalekit-*refs, catalog vs disk,npx skills add . -l) before treating the add as done
Removing Skills
When removing a skill:
- Delete the skill directory from
skills/ - Remove the entry from the Skills Catalog table in
README.md
Renaming Skills
When renaming a skill:
- Rename the directory (must maintain
kit-prefix) - Update the
namefield in the YAML frontmatter - Update the Skills Catalog table in
README.md
Rule Management
Rules live in rules/ as .mdc files. They are always-on guidance (not slash skills). npx skills add does not install them. Keep rules/AGENTS.md as a short index only.
Adding Rules
- Create
rules/kit-<name>.mdcwith YAML frontmatter (description,alwaysApplyand/orglobs) and the rule body - Keep each rule to one concern and under ~50 lines
- Add a row to
rules/AGENTS.md - Update the Rules Catalog in
README.md
Removing or Renaming Rules
- Delete or rename the
.mdcfile - Update
rules/AGENTS.mdand the Rules Catalog inREADME.md
General Guidelines
- No secrets: Never commit credentials, tokens, private hostnames, or work emails
- Keep skills generic: Avoid employer names, internal tools, and proprietary details. Skills must work in any company.
- Maintain catalog: Always update
README.mdwhen changing skills or rules - Installation path: Use
netanelavr/agent-kitin all documentation examples
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.

