(idea): What did I miss cli

[seawatts]

The Problem

Developers often need to catch up on changes in a codebase after being away for a few days or weeks. Manually reviewing git history, reading through commit messages, and understanding the context of changes can be time-consuming and overwhelming. There's a need for a tool that can automatically analyze recent changes and provide a clear, concise summary of what happened in the codebase.

The Idea

Create a CLI tool called "What did I miss" that:

1. Takes a time parameter (e.g., "7 days ago")
2. Fetches git history and changed files for that period
3. Uses an LLM to analyze the changes and generate a human-readable summary
4. Provides insights about:
   • Major features or changes implemented
   • Bug fixes and their impact
   • Architectural changes
   • New dependencies or tools added
   • Breaking changes or migration requirements

The tool would output a well-formatted markdown report that developers can quickly scan to understand what they missed.

Nice to have

• Integration with popular project management tools (Jira, GitHub Issues) to link changes to tickets
• Ability to filter summary by specific directories or file types
• Option to generate different levels of detail (high-level overview vs. detailed changes)
• Support for multiple git repositories in a monorepo setup
• Interactive mode to ask follow-up questions about specific changes
• Customizable LLM prompts for different types of summaries
• Export options (PDF, HTML, etc.)

Audience

Primary audience:

• Software developers who need to catch up after time off
• Team leads who want to stay informed about project progress
• New team members onboarding to an existing project
• Remote team members who need to stay in sync with changes

Reach

This tool could be valuable for:

• Individual developers and small teams
• Large organizations with multiple active repositories
• Open source projects with many contributors
• Remote-first teams that need better change visibility

While this could be a standalone tool, it could also be integrated into existing developer tools or IDEs to increase its reach and adoption.

Level of effort

Low to Medium

Implementation requirements:

• 1-2 engineers
• 1-2 weeks for initial version
• Basic knowledge of git operations and CLI development
• Experience with LLM integration

Experience

Helpful background:

• Python/Node.js for CLI development
• Git API knowledge
• Experience with LLM APIs (OpenAI, Anthropic, etc.)
• Understanding of code review and change management
• Basic NLP concepts for processing commit messages and code changes

Contributing / Open Questions

• What's the best way to handle large repositories with many changes?
• How can we ensure the LLM summaries are accurate and relevant?
• Should we support different LLM providers or stick to one?
• How can we handle sensitive information in commit messages?
• What's the best way to structure the output for different use cases?
• How can we make the tool extensible for different project types?