Automatic Commit Message Generator

What You'll Build

Your AI agent will analyze code changes and generate commit messages following the Conventional Commits specification.

Inputs

Your agent will have access to:

• Git diff: The staged changes in your repository
• Branch name: The current Git branch you're working on
• Recent commits: A list of previous commit messages for context
• Kanban integration: A pre-configured MCP server to fetch ticket details

Tools You Can Use

• Kanban MCP Server: Already integrated into your project. You can fetch the ticket details using the tools provided by this server.

Note: You need to write the code that exposes the tools from the MCP server. See kanban/main.py

Placeholder

Insert the token {git_diff} anywhere in user_prompt.txt - the grading harness will replace it with the actual git diff. Same goes for {branch_name} and {recent_commits}

Requirements

Your generated commit messages must be:

• Concise and Clear: Easy to understand at a glance
• Correctly Formatted: Follow the Conventional Commits specification
• Accurate: Reflect the actual changes in your code
• Consistent: Match the tone and style of previous commits
• Context-Aware: Use ticket information when available (e.g., from branch names like feature/TICKET-123-new-auth-flow)

Example

Here's what a good commit message looks like:

feat(api): add support for pagination

-   Enables cursor-based pagination for large datasets
-   Adds `pageToken` and `limit` query params to endpoints

Evaluation Criteria

Your solution will be evaluated on:

• Correctness: Produces valid Conventional Commit messages
• Relevance: Messages accurately describe the code changes
• Consistency: Matches the style of previous commits
• Tool Integration: Properly uses Kanban server when ticket IDs are present
• Code Quality: Well-structured, readable, and maintainable code