Agent skill
auto-commit
Generates semantic commit messages from staged changes. Analyzes git diff and creates conventional commit messages. Use when users want to commit with good messages, mention "commit", "save changes", or ask for commit message suggestions.
Install this agent skill to your Project
npx add-skill https://github.com/IHKREDDY/agent-skills-ts/tree/main/skills/auto-commit
Metadata
Additional technical details for this skill
- author
- IHKREDDY
- version
- 1.0
- category
- productivity
SKILL.md
Auto-Commit Skill
🎯 Use Case Demo
Scenario
Developer Sarah just finished implementing a new seat selection feature. She has 5 files changed but struggles to write a good commit message.
Before (Manual):
git commit -m "updated stuff" # 😬 Bad commit message
After (With Skill):
# Skill analyzes the diff and suggests:
feat(booking): add passenger seat selection with availability check
- Add SeatSelection model with row and seat properties
- Implement seat availability validation in BookingService
- Add GET /api/seats/{flightId} endpoint
- Update booking flow to include seat assignment
Time Saved: 5-10 minutes per commit × 20 commits/day = 2+ hours/day
When to Use This Skill
- After making changes, before committing
- When unsure how to describe changes
- To maintain consistent commit message format
- For generating detailed commit bodies
Commit Message Format
Following Conventional Commits:
<type>(<scope>): <description>
[optional body]
[optional footer]
Types:
feat- New featurefix- Bug fixdocs- Documentationstyle- Formattingrefactor- Code restructuringtest- Adding testschore- Maintenance
Agent Instructions
When user asks to commit or wants a commit message:
- Get staged changes:
git diff --cached - Analyze the diff to understand:
- What files changed
- What type of change (feature, fix, refactor)
- The scope (which module/component)
- Generate commit message following conventional commits
- Present to user for approval
- Execute commit if approved
Example Prompts
User: "Commit my changes" → Analyze diff, suggest message, commit
User: "What should my commit message be?" → Analyze diff, suggest message only
User: "Commit with message: fix login bug" → Use provided message directly
Demo Script
# 1. Make some changes
echo "// New feature" >> src/feature.ts
# 2. Stage changes
git add .
# 3. Ask agent: "Generate a commit message for my changes"
# 4. Agent analyzes and suggests:
# feat(core): add new feature module
#
# - Add feature.ts with core functionality
# - Implement base feature logic
# 5. User approves, agent commits
Benefits
| Metric | Before | After | Improvement |
|---|---|---|---|
| Commit message quality | Inconsistent | Standardized | ✅ 100% |
| Time per commit | 2-5 min | 10 sec | ⬇️ 90% |
| Changelog generation | Manual | Automated | ✅ Enabled |
| Code review context | Poor | Excellent | ⬆️ 10x |
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
tech-debt-report
Find TODOs, deprecated APIs, and outdated patterns in codebase
sprint-summary
Generate sprint review summary from completed Jira tickets
api-integration
Design and implement REST API integrations with proper error handling, authentication, rate limiting, and testing. Use when building API clients, integrating third-party services, or when users mention API, REST, webhooks, HTTP requests, or service integration.
work-on-ticket
Pulls ticket details from Jira, creates feature branches with proper naming conventions, and handles planning steps. Use when starting work on a Jira ticket, creating branches for tickets, or when users mention "work on ticket", "start ticket", "create branch for", or Jira ticket IDs.
hotfix-workflow
Rapid hotfix branch creation, fix application, and PR to main
create-ticket
Creates Jira tickets with proper formatting, acceptance criteria, and optionally sets up git branches and pull requests. Use when users want to create a new ticket, log a bug, request a feature, or when they mention "create ticket", "new ticket", "log issue", "file bug", or "create Jira".
Didn't find tool you were looking for?