agentsclimarketplace

Commit writer

Skill jordantplows/STARTUP-OS.MD/.claude/skills/commit-writer

Writes a complete conventional commit message from currently staged changes. Analyzes the diff, determines the correct type (feat/fix/ refactor/chore/docs/test), writes a concise summary line, and a full body explaining what changed and why.From its SKILL.md

Install
npx -y skills add jordantplows/STARTUP-OS.MD --skill commit-writer

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
  • runs commandsInstructs the agent to run 8 commands, including `git diff --staged --stat` and 7 more.

SKILL.md

6.5 KB, ~1.5k tokens by cl100k_base, as published. Nobody here has run it

commit-writer

Generates proper conventional commit messages from staged changes.

Instructions

  1. Check what's staged

    git diff --staged --stat
    

    If nothing is staged, inform the user and exit.

  2. Read the actual changes

    git diff --staged
    

    If the diff is very large (>500 lines), sample key files instead:

    git diff --staged --stat
    git diff --staged path/to/key/file.ts
    
  3. Read recent commit history for style

    git log --oneline -10
    git log -1 --format="%B"
    

    This shows the repository's commit message conventions.

  4. Determine the commit type

    • feat: new functionality added
    • fix: bug fix
    • refactor: restructuring without behavior change
    • chore: tooling, config, dependencies, build setup
    • docs: documentation only
    • test: test additions or changes
    • perf: performance improvement
    • style: formatting only, no logic change

    If changes span multiple types, pick the dominant one and mention others in the body.

  5. Write the subject line Format: type: short imperative summary

    • Under 72 characters
    • No period at the end
    • Imperative mood ("add", "fix", "merge" — not "added", "fixed", "merged")
    • Be specific but concise
  6. Decide if body is needed

    Skip the body for:

    • Typo fixes
    • Single small bug fixes
    • Dependency bumps
    • Self-explanatory one-liners

    Write a full body for:

    • Structural changes
    • Multi-file refactors
    • New features
    • Anything that changes behavior
    • Anything where "why" isn't obvious from "what"
  7. When writing a body

    Structure:

    First paragraph: what changed and why, in plain English.
    
    Changes:
    - Group related changes logically
    - Use bullet points for clarity
    - Include file paths or component names
    
    Updated:
    - List modified existing features
    - Explain impact on behavior
    
    Renamed:
    - Old → new mappings
    - Explain why renamed
    

    Breaking changes: Call out explicitly with "BREAKING CHANGE:" on its own line, followed by what breaks and how to migrate.

    User context: If the user provided extra context as an argument to this skill, weave it naturally into the body rather than appending.

  8. Check for unrelated changes

    If staged changes span genuinely unrelated concerns (e.g., a feature plus an unrelated dependency bump), flag this:

    "These look like two unrelated changes:

    1. [feature description]
    2. [dependency/fix description]

    Want me to write one message covering both, or should we split into separate commits?"

    Wait for their answer before proceeding.

  9. Show the message and confirm

    Display the complete commit message in a code block, then ask:

    Use this commit message?
    [Y] yes  [E] edit  [N] cancel
    
  10. On [Y] — commit

    For short messages:

    git commit -m "type: subject" -m "body paragraph"
    

    For complex multi-paragraph bodies, write to a temp file:

    cat > /tmp/commit-msg.txt <<'COMMIT_EOF'
    type: subject
    
    First paragraph explaining what and why.
    
    Changes:
    - item one
    - item two
    
    Co-Authored-By: Claude Sonnet 4.5 (1M context) <[email protected]>
    COMMIT_EOF
    git commit -F /tmp/commit-msg.txt
    rm /tmp/commit-msg.txt
    
  11. On [E] — edit

    Ask: "What should I change?" Wait for feedback, regenerate, show again.

  12. On [N] — cancel

    Confirm cancellation and exit without committing.

Co-Author tag

Always append this line to the commit body:

Co-Authored-By: Claude Sonnet 4.5 (1M context) <[email protected]>

Examples

Example 1: Feature addition

Staged changes: New authentication flow with JWT tokens

Message:

feat: add JWT-based authentication system

Replaces session cookies with JWT tokens for better scalability
and easier mobile client support. Tokens expire after 7 days
and include refresh token rotation.

Changes:
- src/auth/jwt.ts — token generation and verification
- src/middleware/auth.ts — JWT validation middleware
- src/routes/auth.ts — login/logout/refresh endpoints
- package.json — added jsonwebtoken dependency

Updated:
- User model to store refreshTokenVersion for invalidation
- API clients to include Authorization header

BREAKING CHANGE: clients must now send 'Authorization: Bearer <token>'
header instead of relying on cookies. See migration guide in docs/auth.md.

Co-Authored-By: Claude Sonnet 4.5 (1M context) <[email protected]>

Example 2: Bug fix (no body needed)

Staged changes: One-line fix for null pointer

Message:

fix: prevent null pointer in user profile loader

Example 3: Refactor

Staged changes: Reorganized directory structure

Message:

refactor: move executive agents into department directories

Executives now live inside the department they steer instead of
as separate top-level folders. This reflects that an executive
is the steering layer of its org, not a standalone entity.

Changes:
- cfo/ → finance/exec/
- cmo/ → marketing/exec/
- cto/ → engineering/exec/
- cpo/ → product/exec/
- coo/ → operations/exec/
- All department agents updated with reportsTo field

Updated:
- company-os.ts schema — exec state nests under department
- router.ts — agent discovery updated for new structure
- All .md agents — corrected read/write paths

Renamed for clarity:
- product/exec/roadmap.md → product/exec/roadmap-oversight.md
- operations/exec/operations.md → operations/exec/daily-operations.md

No functional behavior change — this is structural only. All
agent logic, watch/reason/act cycles, and company.os coordination
remain the same.

Co-Authored-By: Claude Sonnet 4.5 (1M context) <[email protected]>

Usage

In any conversation with staged changes:

/commit-writer

Or with context:

/commit-writer This fixes the bug reported in issue #42 where users 
couldn't log in after password reset.

The skill will analyze the diff, generate the message, and walk you through confirmation before committing.

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Gives 0 of the 12 instructions most docs writing skills give in ~1.5k tokens

Counted across 1,951 of the 3,904 authors here whose files we hold, read 2026-09-06

  • Use third-person for skill descriptionsin 54 of 1951, across 35 files
  • Start descriptions with Use whenin 43 of 1951, across 29 files
  • Run baseline scenarios before writing any skillin 40 of 1951, across 26 files
  • Use active voicein 40 of 1951, across 36 files
  • Map file responsibilities before defining tasksin 36 of 1951, across 29 files
  • Use checkbox syntax for tracking stepsin 35 of 1951, across 27 files
  • Ask one question at a timein 35 of 1951
  • Offer execution options after saving the planin 33 of 1951, across 24 files
  • Include complete code in every stepin 33 of 1951, across 27 files
  • Design units with clear boundaries and interfacesin 31 of 1951, across 23 files
  • Announce the skill usage at the startin 30 of 1951
  • Verify agent compliance after adding the skillin 29 of 1951, across 17 files

Said here and by no other author read

  • exit if no changes are staged
  • read recent commit history for style
  • determine commit type from standard categories
  • write body for structural or behavioral changes
  • flag unrelated changes for user review
  • append co-authored-by tag to commit body

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 325,949. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.