Overview
Trustworthy, citation-backed answers over your own document library — self-hosted.
An open-source agent for high-accuracy, citation-backed question answering over your PDFs and scientific literature.
PaperQA2 (Future House) is an autonomous retrieval-augmented agent that answers questions over a corpus of documents and returns answers grounded in real citations, reducing hallucination. You run it over your own PDF/library. Free and open source — installation required. Built by Future House and open-sourced; Buy My Agent curates the official link — we do not own or operate it.
Best for: Researchers, analysts and anyone doing literature or document review.
Not for: People wanting an instant hosted chatbot with no install.
You'll know it worked when: You install PaperQA2, index your documents, and get citation-grounded answers.
Capabilities
- ✓Link to the official PaperQA2 (paper-qa) GitHub repository
- ✓Installation + usage instructions
- ✓Bring-your-own-model configuration
Creator-provided example — not verified performance
Requirements
You'll need
- Python + the project's dependencies
- A model provider API key or a local model
- The documents/PDFs you want to query
Intermediate · 30–60 minutes setup
Trust & operating facts
Human-reviewed
Every listing is reviewed by Buy My Agent before it goes live.
What you get & your data
Setup: Connect your tools — some setup or creator handoff is required, as described below.
What you get: the creator's public link that opens this agent on GitHub. You use it in your own GitHub account; the creator can replace the link at any time.
Your chats: they happen in your own GitHub account under GitHub's privacy terms — neither the creator nor Buy My Agent can see your conversations.
Creator data access: The subscriber runs it locally / in their own accounts. Runs on your own machine over your own documents; they stay with you. (creator-declared)
Your contact info: Buy My Agent never gives the creator your email unless you choose to share it.
AI usage: Free and open source. You run it yourself; a commercial model/embedding API can incur that provider's usage charges (a local model avoids them). Point it at documents you have the right to use.
Creator support
Support: No support included (creator-declared)
Questions or a refund request? Message the creator privately — they'll reply to your email.
Reviews
No reviews yet
Subscribe and try it — early reviews help other subscribers decide.