readme-refresh
Skill3 starsChanged 57 days agoMITView on GitHub
Use when the user asks to update, fix, or audit README.md, or when README commands look stale. Checks every command, path, and claim in the README against the actual repo and proposes a minimal diff; never rewrites the whole file unasked.
In this repository7 other files it gives its agents.
- changelogSkill
- claude-md-initSkill
- commit-messageSkill
- dependency-auditSkill
- pr-descriptionSkill
- secret-auditSkill
- test-gap-finderSkill
The file
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 5 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈5
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 9
Read README.md fully. Extract: every fenced command, every relative path/link, every badge/CI claim, install instructions, and the feature …
- Distinctive: line 10
Verify each against reality:
- Distinctive: line 17
If the README lacks a required section (Install, Usage, Test), add a minimal one using verified commands.
- Very specific: line 21
Never delete sections you cannot verify.
- Distinctive: line 22
Keep the author's voice and structure.
- Distinctive: line 23
No new badges, emojis, or headings beyond what is needed to fix a finding.
≈ “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
readme-refresh
Procedure
- Read
README.mdfully. Extract: every fenced command, every relative path/link, every badge/CI claim, install instructions, and the feature list. - Verify each against reality:
- Commands: check the script/target exists (
package.jsonscripts,Makefile,pyproject.toml,Cargo.toml). Dry-run cheap ones (--help,make -n). - Paths/links:
test -e <path>; fordocs/*.mdlinks check the file exists. - Versions: compare stated runtime versions with
.nvmrc,.python-version,engines,Dockerfile. - Features: grep the codebase for the feature's entry point; flag claims with no code behind them.
- Commands: check the script/target exists (
- Produce a findings table:
line | claim | reality | fix. - Apply only the fixes that are factual (wrong command, dead path, wrong version). For structural/marketing changes, propose and wait.
- If the README lacks a required section (Install, Usage, Test), add a minimal one using verified commands.
- Show
git diff README.md.
Rules
- Never delete sections you cannot verify; mark them
<!-- TODO: verify -->instead. - Keep the author's voice and structure. This is an audit, not a rewrite.
- No new badges, emojis, or headings beyond what is needed to fix a finding.
Verify
Maintainers: bash evals/run.sh /tmp/skillkeel-evals readme-refresh from the plugin root runs this case and grades it; records and transcripts sit in evals/readme-refresh/.
evals/readme-refresh/: fixture README claims npm run start and links docs/api.md; repo has npm run dev and no docs/; expected: both flagged and fixed, other prose untouched.
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.