claude-skills
CLAUDE.md0 starsChanged 5 months agoMITView on GitHub
This is the public GitHub repo (robmcquade/claude-skills, MIT licensed) that publishes a curated subset of Claude Code skills for general use. origin is the only remote — no internal mirror, no second push target. Assume every file here, including this one, is world-readable the moment it's committed. - No machine-specific paths. No local drive letters, no absolute filesystem paths, no hostnames or machine names, anywhere in any file. - No private names or personal details. No internal/unreleased project names, no…
In this repository5 other files it gives its agents.
- evaluate-ai-meritsSkill
- improveSkill
- optimize-efficiencySkill
- pressure-testSkill
- set-goalSkill
The file
We split this CLAUDE.md into the rules it gives an agent
4 rules. About 0 are common, 3 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
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
- Distinctive: line 10
No machine-specific paths: No local drive letters, no absolute filesystem paths, no hostnames or machine names, anywhere in any file.
- Distinctive: line 12
No private names or personal details: No internal/unreleased project names, no client or company names, no personal financial or …
- Distinctive: line 14
No secrets: No API keys, tokens, credentials — real or that could be mistaken for real.
- Very specific: line 24
Commit straight to main, push to origin.
≈ “Common” is an estimate: at least 1 in 5 of the 26,980 CLAUDE.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
CLAUDE.md — claude-skills
This is the public GitHub repo (robmcquade/claude-skills, MIT licensed) that publishes a curated
subset of Claude Code skills for general use. origin is the only remote — no internal mirror, no
second push target. Assume every file here, including this one, is world-readable the moment it's
committed.
Hard rules (this repo is public)
- No machine-specific paths. No local drive letters, no absolute filesystem paths, no hostnames or machine names, anywhere in any file.
- No private names or personal details. No internal/unreleased project names, no client or company names, no personal financial or identifying details.
- No secrets. No API keys, tokens, credentials — real or that could be mistaken for real. (The pressure-test test fixtures use well-known published fake test values on purpose — see the README's note on that before treating a scanner hit there as a real leak.)
- Sanitize before publishing. These skills are curated copies of versions maintained privately elsewhere; the private originals evolve independently and are not auto-synced here. When pulling a change over from that private source, strip anything path-, project-, or person-specific first — don't paste it in as-is.
Push model
- Commit straight to
main, push toorigin. No PR gate, no required CI check, no branch protection ritual to route around. - There is no "wrap" step, no work-tracking ledger, and no mirror-publish for this repo — those are conventions from a different, private workflow and don't apply here.
- Normal git hygiene still applies: don't force-push over someone else's commit, keep commit messages honest about what changed.
Before committing
Re-read the diff against the three rules above. If anything looks like it leaked from a private
environment (a path, a name, a credential-shaped string outside the known test fixtures), pull it back
out before it goes to origin.
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.

