Guide · Agent skill
How to use the Caveman Commit Messages agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The caveman-commit skill is a set of instructions that tells your AI agent how to write git commit messages. It produces a short one-line subject in the Conventional Commits format and adds a body only when the reason for a change is not obvious. This guide explains what the skill does, how to install it, how to trigger it, and where it falls short, so you can use it on your own projects.
Caveman Commit MessagesJuliusBrussee/caveman · Apache-2.0 Download the skill Quick start
- 1Download the ZIP. Download the caveman-commit ZIP from this page.
- 2Unzip into a skills folder. Unzip it into ~/.claude/skills/ for all projects, or into the project's .claude/skills/ folder for one project.
- 3Or upload in the Claude apps. In the Claude apps, go to Settings → Capabilities → Skills and upload the skill.
- 4Make your changes. Make and review the code changes you want to commit.
- 5Ask for a commit message. Type /caveman-commit or ask to write a commit message.
- 6Review and paste. Check the code block output and run git commit yourself with the message.
What an agent skill is and what this one changes
An agent skill is a folder of instructions that an AI agent loads when a request matches the skill. This one is named caveman-commit and is defined in a SKILL.md file.
When it is active, the agent writes commit messages in the form type(scope): summary. Allowed types are feat, fix, refactor, perf, docs, test, chore, build, ci, style and revert. The subject should be in the imperative mood, such as add, fix or remove, and aim for 50 characters or fewer with a hard cap of 72. The skill also leaves out filler such as 'this commit does', 'we' or 'now', and leaves out AI attribution.
- Subject line: type(scope): imperative summary, with no trailing period
- Body: only for a non-obvious why, breaking changes, migration notes or linked issues, wrapped at 72 characters
- Always includes a body for breaking changes, security fixes, data migrations and reverts
- References issues at the end, for example Closes #42 or Refs #17
Install the caveman-commit skill
Download the ZIP for this skill from this page. Then choose one of the two options below.
For a personal install that applies to all your projects, unzip the file into ~/.claude/skills/. For a single project, unzip it into that project's .claude/skills/ folder. You can also upload the skill in the Claude apps under Settings → Capabilities → Skills.
- Personal: unzip into ~/.claude/skills/
- Project only: unzip into the project's .claude/skills/ folder
- Claude apps: Settings → Capabilities → Skills, then upload the skill
Trigger the skill with natural requests
You can invoke the skill with the slash command or simply ask in plain language. The skill description lists these triggers: 'write a commit', 'commit message', /commit and /caveman-commit. The README also mentions 'generate commit'.
Ask the agent after you have made your changes, for example by saying write a commit message for these changes. The agent reads the diff and returns the message.
- /caveman-commit
- write a commit
- commit message
- generate commit
/caveman-commitExample: a feature with a reason in the body
Suppose you add a new endpoint for user profiles. A typical verbose message would say that the commit adds a new endpoint. The skill instead keeps the subject short and explains the reason in the body.
The skill outputs the message as a code block, so you can copy it and commit it yourself.
feat(api): add GET /users/:id/profile
Mobile client needs profile data without the full user payload
to reduce LTE bandwidth on cold-launch screens.
Closes #128Example: a breaking API change
Breaking changes always get a body. The skill adds a BREAKING CHANGE note and the migration context that future readers will need.
feat(api)!: rename /v1/orders to /v1/checkout
BREAKING CHANGE: clients on /v1/orders must migrate to /v1/checkout
before 2026-06-01. Old route returns 410 after that date.Safety notes: scripts, network and what it never does
This skill does not run scripts and does not use the network. It does not run git commit, stage files or amend commits, and it does not change any files. It only generates text.
You stay in control of every commit. Review the message, then paste it into git yourself. To return to verbose commit style, tell the agent 'stop caveman-commit' or 'normal mode'.
- Runs scripts: no
- Uses network: no
- Requires nothing listed
- Does not stage, commit or amend
Tips for better commit messages
Give the agent a clear diff or staged changes so the subject describes the actual change. If your project already uses a different convention, check the output before pasting it, because the skill's rules are fixed to Conventional Commits and only partly defer to project style.
If you need a long explanatory body by policy, add the context yourself, because the skill skips the body when the subject seems self-explanatory.
- Stage or show the changes you want described before asking
- Add issue numbers in your request so the agent can include Closes or Refs lines
- Ask for normal mode if you need a verbose message
What people use it for
- Generating a Conventional Commits subject for a bug fix in a team repository
- Writing a body that explains why a non-obvious change was made
- Producing a BREAKING CHANGE message with migration notes for an API rename
- Keeping commit history consistent when several developers use AI agents
Frequently asked questions
Is the caveman-commit skill free?
The source material for this page does not state pricing or licence terms beyond the Apache-2.0 licence listed for the repository. Check the licence file in the upstream project for the terms that apply.
What does caveman-commit require?
The skill lists no requirements. It runs no scripts and uses no network. It needs an AI agent that can load skills, such as Claude, as described in the installation steps.
Does caveman-commit commit my changes for me?
No. The skill only generates the message as a code block. It does not run git commit, stage files or amend commits, so you paste the message and commit yourself.
How do I install caveman-commit?
Download the ZIP from this page and unzip it into ~/.claude/skills/ for personal use or into a project's .claude/skills/ folder. You can also upload it in the Claude apps under Settings → Capabilities → Skills.
Can I use caveman-commit with a different commit style?
The rules are fixed to Conventional Commits. The skill only partly defers to project convention, so it may conflict with a different local style. Review the output before using it on such projects.
How do I turn caveman-commit off?
Tell the agent 'stop caveman-commit' or 'normal mode'. The agent then reverts to a verbose commit style.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use caveman-commit if you want short, consistent Conventional Commits messages and you already work with an AI agent. Pick something else if your project needs a different commit convention, a long body required by policy, or an agent that commits and stages changes for you.
Download the skill