# Commit Discipline

> How to structure commits: atomic, conventional messages, never push to main/master directly.

- **Type:** Skill
- **Install:** `agentstack add skill-eoeboh-agent-skills-commit-discipline`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [EOEboh](https://agentstack.voostack.com/s/eoeboh)
- **Installs:** 0
- **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [EOEboh](https://github.com/EOEboh)
- **Source:** https://github.com/EOEboh/agent-skills/tree/main/skills/commit-discipline

## Install

```sh
agentstack add skill-eoeboh-agent-skills-commit-discipline
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# Commit discipline

Run this whenever committing changes, opening a PR, or when asked to "commit this" /
"push this."

## Rules

- Commits are atomic: one logical change per commit. If a task touched two unrelated
  things (e.g. a bug fix and an unrelated formatting pass), split into separate commits
  rather than bundling them.
- Commit messages follow conventional-commit style: `type(scope): summary`, where type
  is one of feat, fix, refactor, test, docs, chore, style, perf. Summary is imperative
  mood ("add", not "added" or "adds"), under ~72 chars on the first line. Add a body
  only if the "why" isn't obvious from the summary and diff.
- Never commit directly to main or master. If already on main/master, create a branch
  first and say which branch name you're using before committing.
- Never force-push to a shared branch unless explicitly told to.
- Never commit files matching common secret patterns (.env, *.pem, *_key, credentials.*)
  even if they were staged by something else — flag them and stop instead of committing
  silently.
- Never use `git add .` or `git add -A` blindly. Review what's staged (`git status`)
  and add specific files. If genuinely everything changed is meant to be committed,
  say so explicitly in the report rather than defaulting to it.
- After committing, show the commit hash and message, and confirm the branch. Do not
  push unless explicitly asked to.

## Edge cases

- If the working tree has unrelated uncommitted changes from a previous session mixed
  in with the current task's changes: stop and ask which files belong to this commit
  rather than guessing.
- If a repo has its own CONTRIBUTING.md or commit convention documented, defer to that
  over this skill's defaults and say so.
- If a pre-commit hook fails: report the failure verbatim, do not bypass with
  `--no-verify` unless explicitly told to.

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [EOEboh](https://github.com/EOEboh)
- **Source:** [EOEboh/agent-skills](https://github.com/EOEboh/agent-skills)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** yes
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-eoeboh-agent-skills-commit-discipline
- Seller: https://agentstack.voostack.com/s/eoeboh
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.