codegraph-mcpNo-train, on-prem code knowledge graph served to AI agents over MCP — symbols, call edges, cross-language links and blast-radius queries, with a hash-chained audit log of every read.
Why switchBoth build code knowledge graphs for AI agents. codegraph-mcp is a read-side, on-prem MCP service with audit logging; code-graph-rag adds NL→Cypher queries, data-flow tracing and AST-based editing on Memgraph. Compliance-friendly retrieval vs full query-and-rewrite.
Full comparison → codebase-memory-mcpCode intelligence MCP in pure C: tree-sitter knowledge graph over 158 languages, average repo indexed in milliseconds, sub-ms queries, 10x fewer tokens. Single static binary, zero deps.
Why switchSame tree-sitter-graph idea at opposite weights: codebase-memory-mcp is a single static C binary indexing 158 languages in milliseconds for cheap agent queries; code-graph-rag runs Memgraph to buy Cypher, taint edges and structural rewriting. Speed and zero deps vs analysis depth.
Full comparison → cocoindex-codeAST-based semantic code search for coding agents: pipx install, zero config, local embeddings out of the box — a CLI/skill/MCP that cuts agent context ~70% vs grepping. Built on CocoIndex.
Why switchTwo shapes of code retrieval for agents: cocoindex-code does AST-aware semantic search with local embeddings (pipx install, zero infra); code-graph-rag builds an explicit relationship graph you can walk and mutate. Search-and-read → cocoindex; structure queries and edits → this.
Full comparison → GraftContext layer for large codebases: a graph of plain-English markdown nodes — no embeddings, no index — agents read like any repo file. Claude Code hooks + MCP; 42% fewer tokens in its bench.
Why switchBoth are context layers for large codebases, opposite philosophies: graft stores plain-English markdown graph nodes agents read as files (no DB, no embeddings); code-graph-rag stores the actual AST-derived graph in Memgraph. Legibility and zero infra vs queryable structure.
Full comparison →