agentleFS
e-reitbauere-reitbauer/habits/skills/commit/SKILL.md

commit

Skill1 starsChanged 6 days agoMITView on GitHub

Generates and executes Conventional Commits 1.0.0-compliant commit messages from pending changes. THIS SKILL IS MANDATORY FOR EVERY COMMIT REQUEST. Trigger it whenever the user says "commit", "commit this", "commit it", "write a commit", "create a commit", "make a commit", "save this", or invokes /commit — including a bare one-word "commit". Never run `git commit` directly; always route commit requests through this skill. Splits unrelated changes into separate logical commits when appropriate, drafts a message for each, and asks for approval through an AskUserQuestion modal (approve / approve with co-author / reject) before committing, or applies the user's standing choice when they said to always approve one way.

188 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use validate-plan, act-on-feedback and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • shawn-dumasSkill

    validate-plan

    Validate an orchestration plan after pre-flight certification. Runs adversarial review, deep review (verify prompt data against codebase), and PoC gate for risky approaches. Auto-invoked by orchestrate-* skills at Step 8.

    Also in 5 projects that use this

  • MicrosoftSkill

    act-on-feedback

    Act on user feedback attached to the current session. Use when the user submits feedback on the session's changes via the Submit Feedback button.

    Also in 3 projects that use this

  • decocmsSkill

    add-mcp-tools

    Guide for adding new MCP tools with consistent patterns for schemas, tool definitions, registry updates, and Better Auth integration

    Also in 3 projects that use this

  • MicrosoftSkill

    generate-run-commands

    Generate or modify run commands for the current session. Use when the user wants to set up or update run commands that appear in the session's Run button.

    Also in 3 projects that use this

  • mitsuhikoSkill

    oebb-scotty

    Austrian rail travel planner (ÖBB Scotty). Use when planning train journeys in Austria, checking departures/arrivals at stations, or looking for service disruptions. Covers ÖBB trains, S-Bahn, regional trains, and connections to neighboring countries.

    Also in 3 projects that use this

  • mitsuhikoSkill

    pi-share

    Load and parse session transcripts from shittycodingagent.ai/buildwithpi.ai/buildwithpi.com/pi.dev (pi-share) URLs. Fetches gists, decodes embedded session data, and extracts conversation history.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

6 rules. 0 are common, 6 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 168 edited versions we read keep these.

  • 6

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read all pending changes. Run git status --short and git diff, plus git diff --cached for anything already staged, so the full change set …

    line 20
  2. Distinctive:

    If splitting into multiple commits, first show the split plan — one line per planned commit — so the user sees the whole picture before …

    line 27
  3. Distinctive:

    Never commit without going through the AskUserQuestion modal in step 5, unless a standing choice applies (below).

    line 38
  4. Distinctive:

    When the user says to always handle approval one way ("always approve", "always add the co-author", "don't ask me, just commit"), save that …

    line 43
  5. Distinctive:

    Ask with the modal anyway when a commit is unusual: a breaking change, a revert, or files that look unrelated to the task.

    line 51
  6. Distinctive:

    When the user says to ask again, delete the memory.

    line 53

“Common” is measured: kept by at least 6 in 10 of the 168 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Conventional Commit Skill

Generate a Conventional Commits 1.0.0 message from staged changes, confirm with user, then commit.

Workflow

  1. Read all pending changes. Run git status --short and git diff, plus git diff --cached for anything already staged, so the full change set is considered — staged or not.
  2. Group into logical commits. Split when the changes span unrelated concerns, types, or scopes (e.g. a new feature plus an unrelated bugfix → two commits). One concern per commit. If a single file mixes concerns and interactive git add -p is unavailable, keep it in one commit and note the compromise. If everything is one concern, it stays one commit.
  3. Draft a Conventional Commits message for each planned commit (spec below).
  4. If splitting into multiple commits, first show the split plan — one line per planned commit — so the user sees the whole picture before approving.
  5. Approve via modal, per commit. For each planned commit, present its full message, then ask with the AskUserQuestion tool — never plain text. Offer exactly three options:
    • Approve — commit as-is, no attribution footer
    • Approve (Claude co-author) — append the Co-Authored-By: Claude <...> footer
    • Reject — skip this commit; leave its files unstaged, commit nothing for this group
  6. Execute each approved commit: stage only that commit's files (git add <paths>), then git commit -m "$(cat <<'EOF' ... EOF)". Append the co-author footer only when the "Claude co-author" option was chosen (this overrides any global default to always co-author).

Never commit without going through the AskUserQuestion modal in step 5, unless a standing choice applies (below).

Standing choice

When the user says to always handle approval one way ("always approve", "always add the co-author", "don't ask me, just commit"), save that choice to memory: as a project memory when they said it about this project, otherwise for all projects. Check memory before step 5. With a standing choice:

  • Still show the split plan and every message, then commit with the remembered option instead of the modal, and say so in one line ("Committing without co-author, as you asked").
  • The standing choice only replaces the approval modal. It never makes you commit without being asked to commit, and no message is committed without being shown.
  • Ask with the modal anyway when a commit is unusual: a breaking change, a revert, or files that look unrelated to the task.
  • When the user says to ask again, delete the memory.

Commit Message Spec (Conventional Commits 1.0.0)

Format:

<type>[optional scope][optional !]: <description>

[optional body]

[optional footer(s)]

Types:

  • feat — new feature (correlates with MINOR in semver)
  • fix — bug fix (correlates with PATCH in semver)
  • docs — documentation only
  • style — formatting, no logic change
  • refactor — neither fix nor feat
  • perf — performance improvement
  • test — add/fix tests
  • build — build system or external deps
  • ci — CI configuration
  • chore — other non-src/test changes
  • revert — reverts a prior commit

Rules:

  • Description: imperative, present tense ("add" not "added"), lowercase, no trailing period
  • Scope: noun in parens describing affected section — feat(auth):, fix(api):
  • Breaking change: append ! after type/scope — feat(api)!: — AND include BREAKING CHANGE: footer
  • Subject line ≤72 chars
  • Body: blank line after subject, wrap at 72 chars, use - bullets
  • Footer: BREAKING CHANGE: <desc>, Closes #N, Refs #N, Co-authored-by:
  • Multiple footers allowed, one per line

Breaking changes (two required signals):

feat(api)!: rename endpoint

BREAKING CHANGE: /v1/users renamed to /v1/accounts — update all clients

Never include:

  • "This commit...", "I", "we", "now"
  • AI attribution
  • Emoji (unless project convention)
  • Restatement of file names when scope covers it

Examples

Simple fix:

fix(auth): reject expired tokens on refresh

Feature with body:

feat(dashboard): add favorites count badge

Shows total favorited recipes in account header.
Count fetched from /api/favorites/count endpoint.

Closes #42

Breaking change:

feat(api)!: change favorites response shape

BREAKING CHANGE: favorites endpoint now returns { items, total }
instead of a plain array — update all consumers.

Revert:

revert: feat(parser): add async parsing support

Refs: 3c2a1b4

Scope Guidance

Infer scope from changed files:

  • app/pages/ → page name or pages
  • app/assets/store/ → store name or store
  • app/components/ → component name
  • backend/src/ → module name
  • multiple unrelated areas → omit scope

Body: Write Only When

  • Non-obvious WHY behind the change
  • Breaking change details
  • Migration instructions
  • Linked issues/PRs
  • Reverts a prior commit

Skip body when subject is self-explanatory.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.