codebase-memory-mcp: High-performance code intelligence MCP server supporting 162 languages

codebase-memory-mcp is a local MCP server that provides a persistent code knowledge graph for coding agents like Claude Code and Codex. It indexes functions, classes, imports, HTTP routes and their relationships across 162 languages, and adds semantic type resolution for Python, TypeScript, Go, Rust and others via a hybrid LSP. A native C backend exposes 17 MCP tools and supports 45 automatic or conditional client integrations, with indexing and graph queries requiring no hosted service or API key. Reported benchmarks show the Linux kernel (28 million lines) indexing in about 3 minutes, with Cypher relationship queries under 1 ms on an Apple M3 Pro. On token consumption, in the project's five-query comparison, codebase-memory-mcp used about 3,400 tokens versus roughly 412,000 for file-by-file exploration, a gap of about 120x. A preprint study comparing agent performance across 31 repositories reported 10x lower token usage, 2.1x fewer tool calls, and answer quality scores of 83% versus 92%. CBM Lens Beta is a newly added main-branch feature introduced in PR #2068, not yet included in the current v0.11.0 release. The interface offers an IDE-style source reader, local code chat, 3D architecture diagrams and a Galaxy graph, helping developers using coding agents quickly understand unfamiliar codebases, review changes and trace dependencies. Source analysis and graph storage run locally, and users who need this interface must build from source.