Repo Analyzer is an AI coding agent skill designed for in-depth architectural analysis of open-source projects. It targets engineers, researchers, and power users, providing professional-grade reports that detail design insights, trade-off analyses, and generate Mermaid diagrams, enabling rapid evaluation and understanding of project architectures.

How It Works

The tool clones a target repository and initiates a multi-stage analysis. It combines static code scanning with external research, including web searches for reviews and official documentation. An adaptive Q&A phase tailors the analysis based on project specifics. Core modules are analyzed concurrently by parallel subagents, with findings cross-validated across modules and merged with external research into a cohesive Markdown report. This approach allows for a dynamic report structure and deep, nuanced insights beyond superficial code functionality.

Quick Start & Requirements

• Primary Install: npx skills add yzddmr6/repo-analyzer (recommended). Alternatively, manual installation via git clone to ~/.claude/skills/repo-analyzer (macOS/Linux) or %USERPROFILE%.claude\skills\repo-analyzer (Windows).
• Prerequisites: Compatible with AI coding agents supporting the skills format, such as Claude Code, Codex, and OpenClaw.
• Optional Dependencies: For enhanced research capabilities, Tavily MCP (for website crawling) and Brave Search MCP (for alternative web search) can be integrated.
• Links: Plugin Marketplace: /plugin marketplace add yzddmr6/repo-analyzer.

Highlighted Details

• Focuses on architectural rationale ("why") rather than just code functionality ("what").
• Features an adaptive report structure that customizes chapters based on project traits.
• Employs parallel subagent analysis for concurrent deep dives into core modules.
• Includes competitive positioning, comparing design philosophies against similar projects.
• Generates Mermaid diagrams for architecture overviews, data flows, and module sequences.

Maintenance & Community

Contributions are welcomed via issues or pull requests. Core logic resides in skills/repo-analyzer/SKILL.md, with evaluation frameworks in the references/ directory. No specific community channels (e.g., Discord, Slack) are listed.

Licensing & Compatibility

The project is licensed under the MIT License. This permissive license generally allows for commercial use and integration into closed-source projects without significant restrictions.

Limitations & Caveats

Reports are generated in Chinese by default, though the skill attempts to follow the user's input language. Enhanced web research capabilities are dependent on optional MCP server integrations.