Three open-source MCP tools aim to cut AI coding agent token waste on large codebases
A growing category of developer tools is addressing a core limitation of AI coding agents: their inability to efficiently navigate large repositories without repeatedly re-reading files and exhausting context windows. Tools like code-review-graph, Graphify, and codebase-memory-mcp each parse a codebase once using Tree-sitter, store the result as a queryable knowledge graph, and expose it to AI agents via the Model Context Protocol (MCP). code-review-graph (~16k GitHub stars) focuses on blast-radius analysis for pull request reviews, while Graphify (~75k stars, Y Combinator-backed) extends the graph to include documentation, PDFs, and multimedia. codebase-memory-mcp (~22k stars) prioritizes speed and broad language support, claiming up to 99.2% token reduction through hybrid LSP-grade type resolution. All three are MIT-licensed and aim to reduce the API costs and context overhead that slow down AI-assisted code review on mid-to-large projects.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.
Discussion (0)
Log in to join the discussion and vote.
Log in