Omniclip Rag is an MCP server tool developed by Msjsc001, designed to connect "any AI" supporting the Model Context Protocol to a user's private local documents. Built primarily with Python, it incorporates PyTorch and Hugging Face models for its core Retrieval-Augmented Generation capabilities. Distributed as a lightweight Windows package with an optional runtime, Omniclip Rag enables AI clients to access and search a vast array of local file formats, positioning itself as a crucial component for personal knowledge management and AI-powered information retrieval without cloud dependency.

What does Omniclip Rag connect to?

Omniclip Rag functions as an official MCP Server, connecting to "any AI" that supports the Model Context Protocol. It processes and links to a user's private local documents, including Markdown, PDF, and over 1290 other document formats. Internally, it uses SQLite as the source of truth for metadata and LanceDB with bge-m3 for vector retrieval. The server also includes parsers for Logseq, standard Markdown, and Tika-backed formats, making it versatile for diverse local knowledge bases.

Core features and tools

Omniclip Rag provides local-first RAG capabilities for desktop use. It integrates SQLite for metadata and LanceDB with bge-m3 for vector retrieval, supporting a hybrid search mechanism. The server includes parsers for Logseq, standard Markdown, and Tika-backed formats. While specific callable tools are not explicitly detailed, MCP servers offering RAG typically require a tool named rag_search with parameters for search_phrases, an array of 1-5 strings for semantic retrieval.

Who is it for?

Omniclip Rag is for individuals or teams who need to enable AI to search their private, local document collections. This includes users managing personal knowledge bases like Logseq vaults or extensive Markdown file repositories. It offers a solution for integrating AI with local data without sending sensitive information to external services. This is not the right fit if you require cloud-based document processing or do not manage local document collections.

How to get started

The server authenticates using an API key passed as an environment variable. Users are prompted to provide these sensitive values, which are then set in their MCP client configuration and remain on their local machine.

Use cases

• Allowing AI clients to instantly search and retrieve information from a user's local, private Markdown, PDF, and other document formats.
• Indexing and incrementally updating local knowledge bases, such as Logseq vaults and standard Markdown files.
• Providing a hybrid retrieval mechanism that combines lexical search (SQLite FTS5 + LIKE) with semantic vector search (LanceDB + bge-m3) to assemble relevant context for AI tools.

Omniclip Rag offers a powerful solution for integrating AI with local document collections, ensuring privacy and control over data. Its ability to process a vast array of formats and combine lexical with semantic search makes it a strong contender for personal RAG applications. The reliance on environment variables for authentication keeps sensitive data local.