Tool
private-gpt
zylon-ai/private-gpt
What it is
An open-source API server that sits between an application and a local OpenAI-compatible model server, adding retrieval with citations, file ingestion, tool use, MCP connectors and database access.
Who it might suit
Developers building applications on local models who want one API for retrieval, tools and MCP instead of assembling those parts themselves.
What to check first
It does not run models itself, so a separate inference server is needed. Its built-in tools include code execution and web fetch, so check how they are isolated before enabling them, and skills support is marked basic.
How it runs
Self-hosted API server
Written