commit-summary
Skill2 starsChanged 12 days agoMITView on GitHub
Write a commit message body derived from the code behind the change, not from a summary of the diff.
2 other projects use a version of this.
Adopted by
Most recently taken up by
- axetroy/skillstheir own version, first seen Oct 13, 2026
- kennethkeim/skillstheir own version, first seen Jan 24, 2026
Other versions2 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 2
- axetroy/skills★ 1same name
- kennethkeim/skills★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository40 other files it gives its agents.
- AGENTS.mdAGENTS.md
- skills/outsider/references/agents.mdAGENTS.md
- broSkill
- changes-reviewSkill
- ci-auditSkill
- code-cleanupSkill
- code-to-specSkill
- consiliumSkill
- doSkill
- doubtSkill
- editor-configSkill
- explainSkill
- handoffSkill
- issue-analyzeSkill
- issue-flowSkill
- issue-triageSkill
- issue-writerSkill
- labels-syncSkill
- live-probeSkill
- markdown-writingSkill
- next-issueSkill
- npm-releaseSkill
- orchestrateSkill
- outsiderSkill
- package-auditSkill
- pr-reviewSkill
- pr-triageSkill
- react-auditSkill
- reflectSkill
- repo-hardeningSkill
- resolve-conflictsSkill
- security-auditSkill
- session-cleanupSkill
- sharpenSkill
- skill-auditSkill
- solve-issueSkill
- tests-auditSkill
- three-auditSkill
- tsconfig-auditSkill
- working-docsSkill
The file
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 3 are distinctive, and 2 are 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.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 42
Read the actual hunks, not just the stat.
- Very specific: line 62
Read references/commit-body.md and follow it.
- Distinctive: line 79
When the caller hands over design rationale pulled out of source comments — a cleanup pass reporting "Suggested for commit message" — that …
- Very specific: line 114
Check the project's CLAUDE.md, AGENTS.md, or CONTRIBUTING.md for commit conventions.
- Distinctive: line 121
Add only when it applies, after a blank line:
≈ “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
Commit body writer
Produce the part of a commit message that is not recoverable from the diff. Whoever reads the commit later already has the diff; what they do not have is what the running code does that forced the change, which call reaches the broken path, what breaks for someone who updates, when the defect was introduced, and what was noticed and deliberately left alone.
Output the body text only — no subject line, no preamble, no commentary. The caller owns the subject.
Arguments
| Argument | Behavior |
| (empty) | Analyze staged + unstaged changes |
staged only | Analyze the index only |
| a commit ref | Analyze that commit |
| anything else | Treat as additional instructions or context |
Step 1: See the change
git status --short
git diff
git diff --cachedRead the actual hunks, not just the stat. You need the semantics of the edit to weigh it.
Step 2: Weigh the change
Derive (Step 3) when the change alters behaviour, a contract, a public type, or a default, or when it fixes a defect.
Report (Step 4) when the change is mechanical (a rename, a mass import rewrite, a formatting pass), generated (lockfiles, build output, snapshots, regenerated trees), or purely additive scaffolding with no behaviour to describe.
Weight decides, not diff size. A one-line semantic fix derives. A rename across 30 files is one fact and reports.
A mixed change is weighed on its heaviest part: derive for the behaviour change, and give the mechanical remainder one line.
Step 3: Derive
Read references/commit-body.md and follow it. It carries the seven questions, the
table of which code is authoritative for each kind of change, the evidence rules,
the writing rules, and three worked bodies.
Four things from it that decide whether the output is worth its cost:
- Answer against the code, not the diff. Open the implementation behind the declaration, the client behind the handler, the changelog behind the bump.
- Skip any question with no real answer. Three honest paragraphs beat seven padded ones. Padding is the failure mode of this skill.
- Never invent provenance. A hash appears only if a command returned it in this session; an enumerated call path is traced in the source or reproduced. No result means the paragraph is dropped, not softened.
- Plain words, short sentences, in the register of the
explainskill: trace the mechanism on real symbols instead of characterising it. Each paragraph opens on a past-tense verb, and the sentence after it is the reason, stated as behaviour.
When the caller hands over design rationale pulled out of source comments — a cleanup pass reporting "Suggested for commit message" — that text answers why the code is built this way. Fold it into the first paragraph. Do not append it as a block at the end.
Output paragraphs, blank-line separated, wrapped at 80. A change with three facts gets three, not seven:
`stableTypeOrdering` was added for TypeScript 6, where it defaulted to `false`.
TypeScript 7 enables it by default.
`moduleResolution: NodeNext` is implied by `module: NodeNext`, and `esModuleInterop`
defaults to `true` in TypeScript 7.
Kept `strict` explicit so the project adopts any stricter checks added under this
option in future TypeScript versions.Step 4: Report
For mechanical and generated changes, one sentence per line, two or three lines, past tense, no blank lines, no bullets:
Renamed `useToggle` to `useToggleGroup` across 31 call sites.
Regenerated the Codex wrapper symlink trees.Name what a reader cannot see from the file list — the count, the mechanism, the tool that generated it. Do not invent a rationale paragraph for a change that has none.
Repo conventions win
Check the project's CLAUDE.md, AGENTS.md, or CONTRIBUTING.md for commit
conventions. A house style that caps body width, forbids paragraphs, prescribes a
template, or requires another language overrides both formats above. Answer as many
questions as it has room for, in order, and drop the rest.
Footer
Add only when it applies, after a blank line:
Fixes #123/Resolves #456BREAKING CHANGE: <what changed>Co-authored-by: <a human's name> <email>
No attribution
The body ends on its last paragraph: no AI, session-link, or assistant co-author footer, and none copied forward. Callers strip any that slip through at commit time.
Out of scope
Subject lines — the reference examples show one only for context; the caller writes it. Running
git commit, and splitting a change into several commits — Question 7 shows the seam,
the caller decides. Release notes, like PR bodies, are a different contract.
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.

