DEDUPED Hot AI repos. Real use cases. What people build. Get the Monday 5

How to answer questions over your own documents with citations with RAG Re Act Agent

Run a self-hosted knowledge base that answers from uploaded PDF, Word, text, and spreadsheet files.

Short answer: git clone https://github.com/MoMoM101/RAG-ReActAgent.git; then cd RAG-ReActAgent; cp backend/.env.example backend/.env (Windows PowerShell: Copy-Item backend/.env.example backend/.env). All steps.

Credentials. The README publishes a default admin password, so change it right after the first login and before exposing the service.

Who it is forA team lead or ops staff member with internal manuals, policies, or product files who needs answers that point back to the source.

The problemFinding one answer across scattered files is slow, and general chatbots do not cite your documents.

The outcomeUploaded files become searchable, and each answer shows inline citations and source cards.

What you need first

  • Docker and Docker Compose, or Python 3.12 and Node.js 22.22 or newer for a source install
  • An OpenAI-compatible LLM service and an Embedding service, with an API key, for full knowledge Q&A

Steps

  1. git clone https://github.com/MoMoM101/RAG-ReActAgent.git
  2. cd RAG-ReActAgent
  3. cp backend/.env.example backend/.env (Windows PowerShell: Copy-Item backend/.env.example backend/.env)
  4. docker compose --env-file backend/.env -f docker-compose.ghcr.yml up -d
  5. Log in as admin, change the default password, then fill in the LLM and Embedding settings on the settings page.
  6. Upload your .pdf, .docx, .txt, .md, .csv, or .xlsx files and ask a question in the chat.

Prompt for your agent

Clone https://github.com/MoMoM101/RAG-ReActAgent, copy backend/.env.example to backend/.env, and start it with docker compose --env-file backend/.env -f docker-compose.ghcr.yml up -d. Then upload the files in <path-to-document-folder> and ask the question <question>. Expected output: a streamed answer with inline citations and source cards. Verify that the cited file names come from <path-to-document-folder> before reporting the result.

Caveat. Full answers need an LLM and an Embedding service configured in the settings page, and the README's benchmark figures come from a synthetic set that may not match your documents.

Use this in your AI agent

Ask your agent

Using the Deduped MCP, get the setup steps for MoMoM101/RAG-ReActAgent and "Answer questions over your own documents with citations"

Add Deduped to Claude Code once

claude mcp add --transport http deduped https://dedupedai.com/mcp

Cursor, ChatGPT, others

Share: X LinkedIn Hacker News Email Copy link

Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer