Cabinet gives local AI access to your personal document library. Drop in PDFs, CSVs, or web app content, and your AI reads from those files directly instead of guessing from training data.
The project installs via npm and runs entirely on your machine - no cloud accounts, no subscription. It uses RAG (retrieval-augmented generation) - where the AI searches your uploaded documents for relevant text before forming a response, rather than relying solely on what it learned during training. Answers stay grounded in your actual files rather than the AI's general knowledge.
Cabinet works with Claude Code using a bring-your-own-agent model, so you connect it to whatever AI workflow you're already running. It also supports background agents with scheduled "heartbeats" - recurring processes that run on a timer to do things like auto-index new files or run periodic summaries across your document library.
The developer frames it as combining Paperclip-style AI memory with Obsidian-style local storage. The comparison holds: Cabinet keeps everything on your machine while making those files queryable through an AI interface.
The project is early-stage and open source on GitHub. For anyone running Claude Code workflows who wants structured document access - rather than pasting file contents manually into prompts - it fills a real gap. Most commercial RAG tools require a cloud subscription; most self-hosted alternatives require significant configuration. An npm install that gets you a working local knowledge base is a lower-friction option.