claudemem-search

Semantic code search retrieves symbols and functions by meaning rather than exact text matching. This Skill uses AST parsing and embedding-based queries to find relevant code across multi-language repositories, ranked by PageRank to surface the most impactful results first.

Start with the map command to visualize your codebase structure, then use the symbol command to locate specific functions, followed by callers and callees to trace dependency chains. This reveals which code impacts which other code and accelerates change assessment.

Yes. This Skill supports version-aware command fallbacks, so it gracefully adapts to differing claudemem implementations. Commands execute with available APIs, ensuring compatibility across environments without manual reconfiguration.

AST navigation locates exact symbol positions and structural relationships with precision; semantic search retrieves contextually relevant code by meaning. Combined, they enable both targeted exploration and discovery-based comprehension of unfamiliar codebases.

Yes. The Skill applies AST-based analysis and semantic search across multiple programming languages within the same repository, using symbol graphs and PageRank to surface impactful symbols regardless of language.

You need the claudemem dependency installed to build the symbol index and embeddings. The Skill integrates with LanceDB for the index layer, storing AST metadata and semantic vectors that power searches and dependency analysis.