Git Workflow Rules

SkillDev tools

Git workflow rules for this repository. MUST be followed for all git operations, including branching, committing and pushing.

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the Git Workflow Rules skill

What this skill tells your AI

The instructions your AI receives, as published by conceptual-machines/magda-core in .claude/skills/git/SKILL.md and read by ahel’s review.

Creating Branches

The usual base is the current dev/0.X.0 branch. main only moves at release time, so it can trail dev by a long way. Other bases are sometimes right, so check the situation rather than applying a fixed recipe, and ask if it is unclear.

git fetch --prune
git checkout -b <branch-name> origin/dev/0.X.0

NEVER let a feature branch track origin/main. This causes accidental pushes to main.

NEVER push directly to origin/main. All changes go through PRs.

If git status shows third_party/faust and third_party/tracktion_engine modified plus an untracked third_party/JUCE/, the submodules are not actually dirty. The branch is just based on a commit that predates JUCE becoming a first-class submodule. Rebasing onto the intended base clears all three, so do not reset or re-sync the submodules.

Branch Naming

Use the pattern: feat/<short-description> or fix/<short-description>

Examples:

  • feat/midi-cc-pitchbend
  • fix/sampler-load-icon

Committing

  • Only commit when the user explicitly asks
  • Use descriptive commit messages with a short summary line
  • If pre-commit hooks fail (e.g. clang-format), re-stage the formatted files and create a new commit. Do NOT amend
  • Never use --no-verify

Dangerous Commands — NEVER Run Without Explicit User Request

  • git push --force
  • git reset --hard
  • git checkout . / git restore .
  • git clean -f
  • git branch -D

Signals

GitHub stars
206
Forks
23
Last commit
Sep 2026
Advanced
Catalog kind
skill
Gateway key
git-conceptual-machines
Source
github.com/conceptual-machines/magda-core