Claude Context: code search for Claude Code

Claude Context is presented as an MCP server that gives a coding assistant semantic search across a repository. The short describes vector embeddings and hybrid retrieval as the basis for finding relevant code, then explains that the index uses AST-based chunking and incremental updates. The speaker says it can index millions of lines and reduce token usage by 40%, but the supplied material includes no benchmark method or test conditions for that figure.

The accompanying description names Claude Code, Cursor, Gemini CLI and VSCode as clients. It describes Claude Context as open source and running locally, with Merkle-tree indexing that re-embeds only changed files. These details explain the proposed role: retrieving code context for an AI agent when a repository is too large to fit in a model's context window. They do not establish that the agent or its model runs offline.

This is a brief overview rather than a setup guide. It introduces the search and indexing approach but provides no installation commands, hardware requirements or configuration walkthrough. The speaker also calls context retrieval instant; the short does not show timing measurements to support that claim.