Mcp Context Server

Provides persistent multimodal context storage that multiple LLM agents can share through thread-based scoping. You get 13 tools including store_context, search_context, and update_context for managing text and images with metadata filtering, tag-based organization, and date range queries. The real power is in the optional search capabilities: semantic search with embeddings from Ollama, OpenAI, or HuggingFace, full-text search with stemming and boolean queries, or hybrid search combining both with reciprocal rank fusion. Includes automatic LLM-based summarization for search results and cross-encoder reranking for precision. Runs on SQLite by default or PostgreSQL for high-concurrency deployments. Reach for this when you need agents to build and query shared context across conversations without losing state.