The product reduces input token usage for AI coding agents by 94 percent. It indexes a codebase using Tree-sitter and serves relevant chunks via the MCP protocol. It integrates with editors including VS Code, Cursor, Copilot, Codex, ChatGPT, and Kiro.
It uses a hybrid search combining vector similarity and BM25 keyword search with Reciprocal Rank Fusion. It provides session memory so decisions and code areas persist across sessions. It is released under the MIT open source license.
No comments yet.