Semantic, on-demand skill retrieval for Claude Code that saves tokens and improves skill discovery by replacing the native skill listing with vector embedding search.
Graph-based tool retrieval for LLM agents. Builds a tool graph from OpenAPI/MCP specs and retrieves multi-step workflows via hybrid search (BM25 + graph traversal + embedding), recovering accuracy from 12% to 82% with 79% fewer tokens. Also works as an MCP Proxy to aggregate multiple servers behind 3 meta-tools.
Provides context-aware skill selection for AI agents, reducing token usage by 85-98% and improving accuracy through semantic retrieval, session memory, and feedback learning.
MCP proxy server with semantic tool search for LLM coding agents. It reduces context window usage by activating only relevant tools based on user queries.