shared-skills
Skill1 starsChanged 3 days agoMITView on GitHub
Add, edit or remove a skill in the user's shared agent-skills repo, the single source of skills for Claude Code, Codex, Cursor, Pi and Grok. Use when asked to "create a skill", "save this as a skill", "add this to my skills", "update the X skill" or "make this available to all my agents".
In this repository15 other files it gives its agents.
- AGENTS.mdAGENTS.md
- code-auditSkill
- comment-auditSkill
- coordinatorSkill
- correctSkill
- deliverSkill
- handoffSkill
- publishing-a-repoSkill
- releaseSkill
- repo-setupSkill
- test-auditSkill
- triageSkill
- verification-skillSkill
- writing-docsSkill
- writing-readmesSkill
The file
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 2 are distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 16
Run <repo>/bin/skills new <name> (lowercase letters, digits, hyphens).
- Very specific: line 17
Write the frontmatter with only name and description.
- Very specific: line 18
Write the body as direct instructions.
- Distinctive: line 19
Keep the skill harness-neutral: name tools generically ("run the command", "read the file") instead of one harness's tool names.
- Very specific: line 27
Run <repo>/bin/skills status.
- Distinctive: line 29
Commit only when the user asks.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill 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
shared-skills
The user keeps one skills repo that every agent harness on this machine reads. Put personal skills there, not in a harness-specific folder such as ~/.claude/skills or ~/.codex/skills.
Find the repo
Resolve the link: readlink ~/.agents/skills/agent-skills prints <repo>/skills. The repo root is its parent.
Add a skill
- Run
<repo>/bin/skills new <name>(lowercase letters, digits, hyphens). It createsskills/<name>/SKILL.mdand links it. - Write the frontmatter with only
nameanddescription; those two fields work in every harness. The description decides when an agent loads the skill, so state what it does and the phrases that should trigger it. - Write the body as direct instructions. Put scripts and reference files next to
SKILL.mdand refer to them by relative path. - Keep the skill harness-neutral: name tools generically ("run the command", "read the file") instead of one harness's tool names.
Edit or remove a skill
Edit the files in place; every harness reads them through a link, so the change is live. To remove a skill, delete its folder and run <repo>/bin/skills link to prune the link.
Check
Run <repo>/bin/skills status. A CONFLICT or an entry under "same name elsewhere" means an older copy of that skill lives in a harness folder and may shadow the repo copy; tell the user instead of deleting it.
Commit only when the user asks.
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.