Codebase Recon

0•

Traces complex software systems end-to-end to create an exhaustive reconnaissance map before planning architectural changes or refactors.

The Recon skill ensures engineers and AI agents read and thoroughly analyze an existing codebase before drafting plans or modifying code. By mapping entry points, call graphs, state mutations, and contract boundaries across parallel read-only subagents, it prevents hallucinated blast radiuses and undocumented dependencies. Recon integrates seamlessly with knowledge graph tools like codebase-memory or falls back cleanly to search patterns, compiling an actionable markdown Recon Map that catalogs confirmed call sites, seams, failure modes, and unresolved unknowns for bulletproof system refactoring and migration planning.

主要功能

01Integrates with codebase-memory knowledge graphs for high-speed graph traversals with fallback to ripgrep
02Maintains an explicit Unknowns inventory to prevent unverified assumptions during implementation
03Executes concurrent read-only subagents across call paths, state, contracts, and operational failure boundaries
04Generates a structured Recon Map detailing seams, blast radiuses, and explicit file:line locations
050 GitHub stars
06Enforces strict plan discipline by requiring confirmed code references before defining scope

使用场景

01Blast radius scoping and architectural planning for multi-service or cross-module database migrations
02Pre-refactor impact analysis to map out dependent callers, state changes, and API contracts before refactoring legacy modules
03Auditing complex runtime lifecycles and operational failure pathways prior to upgrading critical system infrastructure