search-interceptor

Semantic search reduces token waste by fetching ranked results instead of bulk reads. Before reading 3+ files, pause and run claudemem search to get targeted matches, then read only what's needed.

Semantic search is more efficient than broad globs like **/*.ts or **/*.py when claudemem indexing is available. It ranks results by relevance, cutting unnecessary file reads and token consumption.

Interception pauses bulk file operations before execution, evaluates whether claudemem indexing exists, and recommends semantic search as an alternative. This decision matrix prevents wasteful sequential reads.

Yes. Semantic search intercepts bulk reads in agent workflows, applying to code-analysis tasks, parallelization scenarios, and prompts mentioning file paths. It delivers ranked results with lower token cost.

Semantic search ranks results by relevance and reduces token overhead, while bulk reads process all matches sequentially. Interception evaluates indexing availability to choose the more efficient path automatically.

Yes. The interception protocol evaluates claudemem indexing before recommending semantic search. Without indexing, bulk file reads proceed as normal.