Codebase Research & Documentation
Conducts systematic, hallucination-free codebase exploration and documentation following a strict understand-before-explore methodology.
Codebase Research is a specialized skill for Claude Code that enables deterministic, step-by-step exploration of complex repositories. By enforcing the 'Iron Law' of understanding the problem before touching code, it prevents aimless searching and ensures developers have a deep grasp of system dependencies and data flows. The skill utilizes dedicated agents for file discovery and external context gathering, culminating in a structured research document that serves as a single source of truth for implementation planning.
主な機能
011,240 GitHub stars
02External context gathering via a dedicated web-researcher agent
03Structured data flow tracing to map internal system dependencies
04Automated file-finder agent to identify relevant code paths and reading orders
05Strict 'Iron Law' protocol requiring problem validation before file access
06Automatic generation of timestamped research documents in Markdown format
ユースケース
01Investigating system-wide data flows before refactoring or extending functionality
02Onboarding to a new or complex legacy codebase to understand existing features
03Pre-implementation research to identify all affected dependencies and constraints