Extracts structured business rules and domain logic from existing codebases using automated code intelligence and LLM analysis.
Spec Extract is a specialized capability designed for developers and business analysts who need to reverse-engineer the 'source of truth' within undocumented or legacy codebases. By leveraging advanced symbol clustering and cross-cutting discovery, it identifies decision logic, constraints, and domain behavior that are often buried in implementation details. The skill produces a structured rule catalog that distinguishes intentional business logic from technical infrastructure, making it an essential tool for system audits, codebase onboarding, and preparation for major refactors or rewrites.
主要功能
01Automated symbol clustering using cartog MCP or deep Grep/Glob analysis
02Cross-cutting detection of rules in middleware, database constraints, and environment configs
03Structured rule catalogs with type identification, confidence scores, and source tracking
04Layered analysis that isolates core logic from infrastructure and framework concerns
05Interactive checkpointing to verify discovery scope before deep-dive analysis
060 GitHub stars
使用场景
01Auditing actual system behavior to verify compliance against intended business requirements
02Onboarding new developers to a complex codebase by providing a domain-specific logic index
03Reverse-engineering undocumented legacy business logic for a system rewrite or migration
