OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
RAG & Knowledge Systems

Docpaws

Docpaws is a self-hosted RAG document assistant that indexes PDFs into knowledge bases and answers questions using retrieval scoped to a library, folder or individual file. Its web interface streams responses with source citations, and a configurable retrieval-distance threshold can trigger refusals when retrieved material is too weak.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Docpaws — OpenViST project profile in RAG & Knowledge Systems.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Docpaws when you need a self-hosted PDF question-answering workflow with retrieval scope controls, citations and configurable refusal behavior, and can manage its setup and model credentials.

Who it is for

  • Individuals or teams who want to build a searchable assistant over their own PDF collections.
  • Developers comfortable configuring a Python backend, web frontend, model credentials and supporting services.

Documented uses

  • Upload PDFs and organize them into knowledge bases and folders.
  • Ask questions across a full library or limit retrieval to a folder or single file.
  • Review streaming answers alongside citations to retrieved sources.
  • Use incremental indexing as documents are added.
  • Run the included Golden 20 evaluation for repeatable RAG response checks.
Before you choose

Requirements & limits

The repository is public and identifies the project as MIT-licensed. The documented local setup requires LLM and embedding API credentials.

  • Docpaws is self-hosted; the documented setup requires configuring and running the application.
  • The stack separates the FastAPI/SQLModel backend from the Vue 3/Vite frontend, with FAISS for indexing.
  • Docker Compose can provide PostgreSQL, Redis and MinIO; SQLite is described as the default when DATABASE_URL is unset.
  • PDF indexing is a separate step, with the README describing an optional Celery worker for processing uploads.
  • Retrieval scope and a distance threshold offer controls over which documents inform answers and when the system refuses.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Set up Docpaws for local development

Start the documented backend and web frontend locally, with the project’s optional Docker-managed services.

Before you start

  • A local copy of the Docpaws repository.
  • Conda or Anaconda with Python 3.11 for the backend.
  • Node.js 20+ for the frontend.
  • LLM and embedding API credentials.
  • Docker if using the PostgreSQL, Redis and MinIO services provided through Docker Compose.

Steps

  1. From the repository root, run the command below if you want the included PostgreSQL, Redis and MinIO services.

    Documented command
    docker compose up -d

  2. Create and activate a Conda environment with Python 3.11, then install backend dependencies with the command below from backend/.

    Documented command
    pip install -r requirements.txt

  3. Copy backend/.env.example to backend/.env and fill in the required API credentials and any other needed settings.

  4. From backend/, start the API with uvicorn docpaws.main:app --reload --port 8000.

  5. From frontend/, run the command below and the command below to start the web interface.

    Documented command
    npm install
    Documented command
    npm run dev

Expected outcome

The README’s local development setup provides a backend at port 8000 and a frontend at port 3000, with Vite proxying `/api` to the backend.

Guide limitations

  • PDF uploads require indexing; the README describes a separate Celery worker and a configured CELERY_BROKER_URL for that stage.
  • This guide covers the documented local development setup, not a production deployment.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
13 October 2026
Original website title
biao994/DocPaws · Official homepage title · retained in its source language
How we review sourcesSuggest a correction