privacy. A self hosted SearXNG still forwards search queries to upstream engines on the internet.
Who it is forA researcher who wants cited reports without sending prompts to a hosted AI service.
The problemHosted research tools see every question you ask.
The outcomeA web UI on port 5000 that runs research with a local model and cites sources.
What you need first
- Docker, or Python with pip
- Ollama or another OpenAI compatible LLM endpoint
- SearXNG for web search
Steps
- Install Docker
- Run
curl -O https://raw.githubusercontent.com/LearningCircuit/local-deep-research/main/docker-compose.yml && docker compose up -d - Open the web UI and create your account
- Ask a research question and read the cited report
Prompt for your agent
Using LearningCircuit/local-deep-research, run `curl -O https://raw.githubusercontent.com/LearningCircuit/local-deep-research/main/docker-compose.yml && docker compose up -d`, then research <my question> and save the cited report.
Caveat. A GPU makes local models much faster; there is a GPU override compose file.
Use this in your AI agent
Ask your agent
Using the Deduped MCP, get the setup steps for LearningCircuit/local-deep-research and "Run a private research assistant with Docker Compose"
Add Deduped to Claude Code once
claude mcp add --transport http deduped https://dedupedai.com/mcp
Every Monday: the 5 hottest repos and exactly how to use them. Free, one email a week, unsubscribe anytime.
Get them freeShare: X LinkedIn Hacker News Email Copy link
Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer