Awesome MCP › Other MCP Servers

m1rl0k/Context-Engine

⭐ 3 Python added to this list on 2026-03-01 repository created 2025-10-08

Context-Engine provides an open-core, self-improving code search solution that integrates with AI coding assistants. It leverages a hybrid search approach combining dense and lexical methods with a reranker, alongside ReFRAG micro-chunking for efficient context retrieval. The platform supports local LLM prompt enhancement and offers dual SSE/RMCP endpoints. The project streamlines deployment with a single command (`ctx quickstart`) that sets up services, indexes codebases, and warms up models, utilizing Qdrant for vector storage. It's designed to integrate seamlessly with various MCP-compatible clients like Cursor, Windsurf, Roo, Cline, and Codex. Context-Engine also includes agent skills for popular AI coding assistants such as Claude, Codex, Gemini, and Augment, teaching them to prioritize Context-Engine's MCP tools for code exploration over generic search methods. Key features include built-in authentication, a unified MCP endpoint for indexing and memory services, and automatic collection injection for workspace-aware queries. Its architecture for local mode emphasizes zero authentication and single-tenancy, making it suitable for personal use, with components like embedding services (ONNX), an LLM decoder (llama.cpp), a learning worker for adaptive reranking, and a file watcher for auto-reindexing. Storage is handled by Qdrant for vector data, Redis for caching, and direct filesystem access.

https://github.com/m1rl0k/Context-Engine

AIMCPModel Context Protocolcode searchLLMretrieval augmented generationRAGQdrantAI coding assistantscontextembeddingindexingReFRAGhybrid searchCLIVS Code extensionClaudeCodexGeminiAugment