Enables high-speed, exact-match keyword searching within the agent-memory system using BM25 scoring.
The BM25 Keyword Search skill provides a powerful full-text search capability for the agent-memory system, allowing developers to locate specific function names, error messages, and technical identifiers that semantic search might overlook. Built on the high-performance Tantivy index, it offers lightning-fast retrieval with customizable scoring, phrase matching, and robust index management tools like rebuilding and automated pruning. This skill is an essential component of a hybrid retrieval strategy, ensuring precision when navigating large codebases or technical logs where exact terminology is more critical than conceptual similarity.
주요 기능
014 GitHub stars
02Support for exact phrase matching and prefix-based queries
03Comprehensive index lifecycle management including pruning and optimization
04Real-time index statistics and status monitoring via CLI
05Fast full-text search powered by the Tantivy BM25 indexing engine
06Seamless integration with the agent-memory retrieval tier system
사용 사례
01Filtering memory results when semantic vector search returns too many irrelevant matches
02Troubleshooting system issues by searching for exact error strings
03Locating specific function or variable definitions across a project
