---

name: smart-commit description: Self-evolving commit message generator that creates conventional commit messages and learns from your feedback to improve over time. Use this skill when you need to generate git commit messages following conventional commit format (type: description). This skill tracks performance, learns project-specific patterns, and automatically evolves to generate better messages based on your feedback. license: MIT

Smart Commit - Self-Evolving Commit Message Generator

Creates conventional commit messages that evolve based on your project's patterns and feedback.

Overview

This skill generates conventional commit messages by analyzing git diff, learns your project's specific commit patterns, and adapts message style based on your feedback to improve over time.

Current Version: 1.0.0

Self-Evolution Enabled: This skill learns from your feedback to improve commit messages over time.

When to Use This Skill

Use this skill when you want to:

• Generate conventional commit messages from git diff
• Get consistent commit message formatting across your project
• Have commit messages that improve based on your feedback
• Track commit message quality over time

Instructions

Generating Commit Messages

When invoked, analyze the git diff and create a conventional commit message following these rules:

Format: <type>(<scope>): <description>

Types: feat, fix, docs, style, refactor, test, chore, perf, ci, build, revert

Guidelines:

• Use imperative mood ("add" not "added" or "adds")
• Keep description under 72 characters
• Don't end with a period
• Include scope for specific modules/components
• Reference issues in footer if applicable

Examples:

• feat(api): add user authentication endpoint
• fix(ui): resolve button alignment on mobile
• docs(readme): update installation instructions

After generating the message, ask the user: "Is this commit message satisfactory? (yes/no/edit)"

Recording Feedback

After user response, record the feedback for learning:

Copyconst engine = require('../../lib/evolution-engine');
await engine.recordExecution('smart-commit', {
  success: userResponse === 'yes',
  execution_time: elapsedTime,
  input: { diff: gitDiff },
  output: { message: commitMessage },
  feedback: userFeedback
});

Performance Tracking

This skill tracks:

• Success Rate: Percentage of accepted messages
• Common Types: Most used commit types for this project
• Typical Scopes: Frequently used scopes
• Average Length: Typical message length
• Feedback Patterns: What users like/dislike

Evolution Metadata

Current performance tracked in meta.json:

Copy{
  "version": "1.0.0",
  "metrics": {
    "success_rate": 0.0,
    "total_executions": 0,
    "acceptance_count": 0,
    "rejection_count": 0
  },
  "patterns": {
    "common_types": [],
    "common_scopes": [],
    "project_specific_rules": []
  }
}

Evolution Log

Usage Example

Copy# User: /commit

[Skill analyzes git diff]

# Skill: Generated commit message:
# feat(api): add user authentication endpoint

# Skill: Is this commit message satisfactory? (yes/no/edit)

# User: yes

[Skill records success and updates metrics]

Configuration

Evolution settings in meta.json:

Copy{
  "evolution_config": {
    "min_samples_before_evolution": 20,
    "min_improvement_threshold": 0.1,
    "auto_evolve_enabled": true
  }
}

Variants

Previous versions are stored in variants/ directory for reference and rollback.

References

• ../../lib/evolution-engine.js - Core evolution engine
• ../evolution-engine/SKILL.md - Evolution engine documentation