comprehend-code

# Comprehend-Code (代码理解) Read, understand, and explain code with structured output. ## When to Use - "Read this Python file and explain the class hierarchy" - "What does this function do?" - "Give me an overview of this codebase" - "Explain the architecture of this project" - "What design patterns are used in this code?" ## When Not to Use - Document understanding (specs, contracts, manuals) → use comprehend-doc - Debugging specific errors → use debugging skill - Quick code formatting questions → not needed ## Procedure 1. **Identify context** — detect programming language, framework, runtime, file type. 2. **Find entry points** — locate `main()`, handlers, route registrations, or script entry points. 3. **Map data structures** — identify key types, interfaces, classes, and their relationships. 4. **Trace data flow** — follow input → processing → output through the code paths. 5. **Map dependencies** — list external packages, imports, API calls, databases. 6. **Synthesize intent** — explain what the code is designed to do, not just what it does line by line. ## Pitfalls - **Cross-language misjudgment**: Do not assume the same patterns apply across languages (e.g., Python duck typing vs Java interfaces). Always state the language first. - **Excessive nesting expansion**: Stop at 3 levels deep unless the user specifically asks for deeper detail. Full call tree expansion drowns signal in noise. - **Framework-specific assumptions**: Do not infer framework conventions wit...