rlm-search

What problem does it solve?

Locating relevant code, documentation, and architecture context in large repositories is slow and error-prone. This skill enforces a deterministic three-phase search workflow that starts with a lightweight summary, proceeds to a semantically informed vector search, and ends with a precise grep/exact-match pass, dramatically reducing toil and missed context.

Core Features & Use Cases

• Three-phase search pipeline: Phase 1 performs a rapid RLM Summary Scan, Phase 2 retrieves semantically relevant chunks via vector search, Phase 3 narrows to exact matches with targeted grep.
• Scoped, reproducible results: Always works in a bounded, auditable sequence to prevent missed context and ensure repeatable results across large codebases.
• Use Case: When locating architecture docs or the implementation details of a function across a sprawling project, use this skill to surface the most relevant material quickly and with high precision.

Quick Start

Invoke the rlm-search to locate architecture or code references by starting with a summary scan, followed by vector search, and finishing with a scoped exact match.