code-research-routing

To investigate code architecture and data-flow patterns in a large codebase, this Skill routes queries to ChunkHound for semantic analysis, mapping dependencies and structures efficiently.

The best way to search for specific strings and file patterns is using native tool fallbacks like Grep, Glob, and Read, which the Skill routes to for direct string matching tasks.

Yes, you need MCP connectivity to use semantic code research routing, as the framework requires ChunkHound integration and Native Tools to be available via MCP for proper query execution.

The decision framework chooses between semantic search and native search tools by analyzing query characteristics, routing high-level architecture questions to ChunkHound and direct string queries to Grep or Glob.

You should avoid using semantic routing for codebase investigations when performing direct file searches or exact string matches, as the framework falls back to native tools like Grep and Read for these specific tasks.