소개
The Code Search Selector acts as an intelligent mandatory gate for all codebase investigations, preventing 'tool familiarity bias' where AI models default to inefficient grep or bulk file reads. By intercepting keywords like 'audit', 'trace', and 'investigate', it directs the workflow toward semantic search via claudemem for conceptual understanding, or reserved grep for exact matches. This skill significantly reduces token waste and improves accuracy when mapping architectures, understanding data flows, or conducting comprehensive code audits by ensuring the right tool is used for the specific search intent.
