pdf-brain
pdf-brain is a local, agent-first knowledge base that indexes PDF and Markdown files with semantic and full-text search, AI-generated titles, summaries and tags, and an MCP server for use with Claude, Cursor and other AI assistants.
pdf-brain indexes PDF and Markdown files into one searchable local knowledge base, using Ollama for embeddings and an optional local LLM to extract titles, summaries, tags and concepts, organized with an SKOS-style hierarchical taxonomy. Search combines vector similarity with full-text search over the same index.
It is agent-first: by default it emits a single JSON envelope to stdout for programmatic use, with a capabilities command exposing its full command and tool contract as JSON Schemas, and a text-format option for a human-readable, TUI-rendered view instead. It runs as an MCP server so AI assistants such as Claude and Cursor can query the knowledge base directly.
It installs as a standalone binary or via npm, runs entirely on the user's own machine with no API costs, and is written in TypeScript under the MIT license.