This MCP server, developed by Jean Ibarz, provides tools for listing and retrieving content from different knowledge bases. Built with TypeScript and leveraging libraries like Faiss and Hugging Face, it offers functionality to access and search through structured knowledge repositories. The implementation focuses on efficient retrieval and vectorization of knowledge base content, enabling AI models to query and utilize diverse information sources. By connecting AI systems with organized knowledge stores, this server facilitates use cases such as question answering, information retrieval, and knowledge-based reasoning, making it particularly useful for applications requiring access to curated information across multiple domains.
No reviews yet. Be the first to review!
Sign in to join the conversation
Highlight AILists the available knowledge bases.
Retrieves similar chunks from the knowledge base based on a query. Optionally, if a knowledge base is specified, only that one is searched; otherwise, all available knowledge bases are considered. By default, at most 10 document chunks are returned with a score below a threshold of 2. A different threshold can optionally be provided using the 'threshold' parameter.
Retrieves KB snippets and asks a configured local/OpenAI-compatible LLM to answer with citations.
Lists registered embedding models and the active model.
Returns read-only corpus, index, model, cache, and transport statistics.
Compares retrieval results across two persisted FAISS index versions.
Writes a text document into a KB through the guarded MCP mutation path.
Deletes a KB-relative document through the guarded MCP mutation path.
Forces a global FAISS rebuild, optionally validating a named KB before the rebuild.
