Provides a memory and code intelligence layer for software agents to enhance AI-assisted development.
Sponsored
LexRAG is an MCP-native server designed to provide memory and code intelligence for LLM agents, transforming code repositories into queryable graphs. It addresses common RAG agent limitations by offering persistent memory across sessions, efficient context management, and cross-file reasoning capabilities. This enables agents to answer complex architectural, impact, and planning questions without repeatedly processing the entire codebase, thereby saving token budgets and improving development speed.
Key Features
01Code graph intelligence for cross-file dependency answers
02Impact-scoped test selection for smarter test runs
031 GitHub stars
04Persistent agent memory for decisions and episodes across sessions
05Hybrid retrieval (vector + BM25 + graph) for enhanced relevance
06Temporal model for historical queries and change diffs
Use Cases
01Understand a subsystem by querying dependency maps and concrete code slices
02Coordinate multiple agents with ownership, task context, and durable progress tracking
03Plan safe changes by analyzing impact and running focused tests
