Claude Context
An MCP plugin that adds semantic code search over an entire codebase to Claude Code and other AI coding agents, storing embeddings in a vector database instead of loading whole directories into context.
Claude Context is a Model Context Protocol plugin that adds semantic code search to Claude Code and other AI coding agents, so an agent can search for relevant code across millions of lines without a multi-round manual discovery process, and bring the results straight into its context window.
Rather than loading entire directories into a model's context for every request, which becomes expensive on a large codebase, it indexes the codebase into a vector database and retrieves only the code related to the current request, keeping token costs manageable as a project grows.
Setting it up requires a vector database, using either a self-hosted instance or a free tier on Zilliz Cloud, and an OpenAI API key for the embedding model. For Claude Code, it is added as an MCP server through the command line; configuration for other MCP clients, such as the OpenAI Codex CLI, is also documented.