DeepFetch acts as a powerful Model Context Protocol (MCP) server, designed to enhance AI agent capabilities with rich, evidence-based web search and PDF content extraction. Instead of simple link lists, it delivers reranked snippets from unique domains, ensuring agents receive highly relevant information. It seamlessly integrates Kagi for discovery, Scrapfly for content extraction, and local ONNX for semantic reranking, along with dedicated handling for PDF documents. Compatible with popular clients like Claude Desktop, Gemini CLI, and Codex CLI, DeepFetch simplifies deployment via Docker and offers a clean upgrade path for managed environments.
Key Features
01Semantic web search with evidence-rich snippet reranking
02Simplified Docker deployment with Kagi and Scrapfly integration
03First-class PDF extraction and content retrieval
04Local ONNX reranking for contextual relevance
05Compatibility with Claude Desktop, Gemini CLI, and Codex CLI (MCP clients)
060 GitHub stars
Use Cases
01Time-sensitive factual lookup and current event research for AI agents
02Extracting text and page-numbered matches from reports, filings, or manuals
03Public web research requiring source-backed evidence
