ruchernchong/commit

Pre-Commit Security Check

Before committing, ensure GitLeaks is configured in the project:

1. Check for Husky setup: Look for .husky/pre-commit
2. Verify GitLeaks integration: Check if gitleaks protect is in the pre-commit hook
3. Auto-configure if missing:
   • If .husky/ exists but GitLeaks is missing, add gitleaks protect --staged --verbose before any lint-staged command
   • If .husky/ doesn't exist, run npx husky init first, then configure GitLeaks

Example .husky/pre-commit with GitLeaks:

#!/usr/bin/env sh
. "$(dirname -- "$0")/_/husky.sh"

# Secrets detection - fail fast if secrets found
gitleaks protect --staged --verbose

# Lint staged files (if present)
npx lint-staged

Only proceed with the commit after confirming GitLeaks is properly configured.

Language Conventions

Infer language style from the project:

• Analyze existing commit messages, documentation, and code comments to detect the project's language variant (US English, UK English, etc.)
• Match the spelling conventions found in the project (e.g., "organize" vs "organise", "color" vs "colour")
• Maintain consistency with the project's established language style throughout commit messages

Create git commits with a balanced approach - keep related changes together, split only when potentially huge:

COMMIT MESSAGE RULE: ALWAYS use short messages (max 50 characters)

1. Pull latest changes from remote to ensure branch is up to date (git pull)
2. Show current git status and analyse all changes
3. Check conversation context for GitHub issue references:
   • If a GitHub issue is mentioned in the conversation, determine if changes close the issue (complete implementation) or relate to it (partial work)
   • Add appropriate footer: "Closes #123" or "Relates to #123"
   • If no issue is mentioned in context, proceed without issue reference
4. Assess the scope of changes:
   • Small to medium changes: Keep related changes in a single commit
   • Large changes: Only split when the changeset is potentially huge and mixing unrelated functionality
5. For normal commits:
   • Stage all related changes together
   • Create ONE short, descriptive commit message (max 50 characters)
   • Focus on the main purpose of the change
   • Add GitHub issue footer if applicable (from step 3)
6. For huge changesets only:
   • Group by major functional areas
   • Stage files by logical groups
   • Create separate commits for distinct features/fixes

Commit Message Requirements (ALWAYS ENFORCE):

• Maximum 50 characters - no exceptions
• Use present tense verbs (add, fix, update, remove, refactor)
• Be specific but concise (e.g., "fix auth redirect bug", "add user search")
• Prefer cohesive commits over artificially split ones

This approach maintains clean git history with consistently short, readable commit messages.