🔥 Hot Repo: 10K Stars — MCP That Cuts Claude Code Costs 40%

One-liner — Claude Context is an MCP plugin that indexes your entire codebase into a vector database so Claude Code and other AI agents retrieve only relevant code, cutting prompt costs by 40%. - Repo: zilliztech/claude-contexthttps://github.com/zilliztech/claude-context - Stars: ⭐ 10.9k +871 today - Language: TypeScript - License: MIT --- What It Does Claude Context adds semantic code search to Claude Code, Cursor, Gemini CLI, Codex CLI, and any MCP-compatible agent. It indexes your repository into a Milvus/Zilliz Cloud vector database using AST-aware chunking, then serves only the relevant snippets on demand. The retrieval combines BM25 keyword matching with dense vector embeddings — a hybrid approach that outperforms either method alone on precision and recall. Why It's Blowing Up The problem is universal: large codebases break AI coding assistants. Loading a 100k-line monorepo into context is either impossible context window exceeded or prohibitively expensive token cost. Claude Context sidesteps both by treating the codebase as a persistent, queryable knowledge base — the agent fetches what it needs, nothing more. Zilliz's own benchmarks show a 40% token reduction under equivalent retrieval quality, and that's the number developers are testing and sharing. The plugin hit 1 on GitHub trending in early May and crossed 10k stars since launch, helped by a dead-simple one-command install that works without any agent-side code changes. The timing is also right: Anthropic's MCP standard is now supported by every major AI coding assistant Cursor, Windsurf, VS Code, Gemini CLI, Qwen Code, Cline, turning claude-context into a universal plugin rather than a Claude-only tool. Zilliz, the company behind the Milvus vector database, brings production-grade infrastructure credibility that most hobbyist MCP projects lack. Key Features - Hybrid search — combines BM25 keyword matching with dense vector embeddings for higher recall than either alone - Incremental ind