Commit

Purpose: Create a clean, well-documented commit Mode: Git operations with user confirmation required Usage: /commit [scope flags]

Iron Laws

1. •NEVER COMMIT WITHOUT EXPLICIT APPROVAL — Valid approval per ai-assistant-protocol, plus commit as a domain-specific term. Silence, questions, "okay" are NOT approval.
2. •NEVER COMMIT SECRETS — .env, credentials, API keys. Scan before staging.
3. •ONE CONCERN PER COMMIT — If changes include both a feature and a refactor, suggest splitting into separate commits.

Constraints

• •Read + git only — Do not modify source code
• •Never force push without explicit request
• •Never amend commits you didn't create
• •Never skip hooks without explicit request

Scope Flags

Note: Command examples use npm as default. Adapt to the project's package manager per ai-assistant-protocol — Project Commands.

Workflow

Step 0: Branch Safety

CURRENT_BRANCH=$(git branch --show-current)
git status --porcelain
MAIN=$(git symbolic-ref refs/remotes/origin/HEAD 2>/dev/null | sed 's@^refs/remotes/origin/@@' || echo "main")

If on main/master: Warn and suggest creating a feature branch. Wait for response.

Step 1: Review Changes

git diff $MAIN...HEAD -- [scope-paths]
git diff -- [scope-paths]
git diff --staged -- [scope-paths]

## Changes to Commit

**Branch:** `[current branch]`

**Modified:** `path/to/file.ts` — [brief description]
**Added:** `path/to/new.ts` — [purpose]
**Deleted:** `path/to/old.ts` — [reason]

**Stats:** X files changed, +Y insertions, -Z deletions

Step 2: Mixed-Concern Check

If changes include different concern types (feature + refactor, or feature + config), flag it:

These changes appear to mix concerns:
- **Feature:** [files related to new behavior]
- **Refactor:** [files with structural changes only]

Split into separate commits? (yes / no)

Step 3: Security Scan (Always Runs)

Scan changed files for security issues before committing:

# Secrets detection in changed files
grep -rn --include="*.ts" --include="*.tsx" --include="*.js" --include="*.json" \
  -E "(api[_-]?key|secret|password|token|credential|private[_-]?key)\s*[:=]" [scope-paths]

# Insecure patterns
grep -rn --include="*.ts" --include="*.tsx" --include="*.js" \
  -E "(eval\(|new Function\(|innerHTML\s*=|dangerouslySetInnerHTML|\.exec\(|rejectUnauthorized:\s*false)" [scope-paths]

If secrets detected: STOP. Warn the user. Do NOT proceed to commit. If insecure patterns detected: Flag for review — ask user to confirm these are intentional before proceeding.

Exclude test files and example/documentation files from blocking — flag them as informational only.

Step 4: Validate (Optional)

npm run typecheck
npm run lint
npm run test -- [affected]

Step 5: Confirm

**Suggested commit message:**

type: [description]

[optional body]

Options: **yes** / **edit** / **review** / **cancel**

GATE: Do NOT run git commit until user responds with explicit approval.

Step 6: Commit

git add [scope-paths]  # NOT -A unless scope is "all"
git commit -m "[message]"

Step 7: Report

**Committed:** `abc1234` — [type](scope): [description]
**Files:** X changed

**Next:** Push? Create PR? Continue working?

Commit Message Format

[type](scope): [short description]

[optional body]

[optional footer: references, breaking changes]

Types

Rules

• •Imperative mood ("add" not "added")
• •Max 50 characters subject, 72 body
• •No period at end
• •Every message answers: what changed and why

Banned Messages

"update code", "fix bug", "changes", "misc", "wip", "stuff", "updates"

Issue References

Fixes #123 / Closes #123 (closes on merge) — Refs #123 (links without closing)

AI Attribution

When AI wrote most of the code: Co-Authored-By: Claude <noreply@anthropic.com>