Local Deep Researcher

Use it when

• You need iterative multi-cycle web research on a topic, with gap analysis and follow-up queries, condensed into one cited markdown summary instead of manually running and pasting searches
• You want a research assistant that uses a local LLM via Ollama or LMStudio so no LLM API key or per-token cost is involved

What it solves

Not the fit when

• Research over private local documents (it searches the public web, not your files)
• Fully offline use (web search via DuckDuckGo, SearXNG, Tavily, or Perplexity still requires internet)
• Production multi-user research APIs without additional deployment work
• doc-cloud-upload-risk
• embedding-recall-poor

Install

git clone https://github.com/langchain-ai/local-deep-researcher.git && cd local-deep-researcher && cp .env.example .env (set LLM_PROVIDER=ollama or lmstudio), then launch: uvx --refresh --from "langgraph-cli[inmem]" --with-editable . --python 3.11 langgraph dev

Invoke

Open the LangGraph Studio Web UI at the printed URL, set the model in the configuration tab, and give the assistant a research topic; the graph outputs a markdown summary with source citations

Alternatives

No reviewed alternatives recorded yet.