GPT Researcher 🧩 Workflow Open source

What it is

GPT Researcher by Assaf Elovic (Apache-2.0) is an open-source deep-research agent. A planner splits your question into sub-questions, crawler agents search and read sources in parallel, and a writer compiles a factual report with citations, typically several thousand words. It can research the web, your local documents or both, and export to Markdown, PDF or Word. It also ships as a Python package, a web app, an MCP server and an Agent Skill.

Key features

• Planner and parallel execution agents for faster research
• Cited reports exported to Markdown, PDF and DOCX
• Web, local-document and hybrid research modes
• Deep Research mode with recursive exploration
• Works with OpenAI, Anthropic, Gemini, Ollama and other LLMs
• Python package, FastAPI/Next.js app, MCP server and Agent Skill

How to install

Install the library with pip install gpt-researcher, or clone the repo, pip install -r requirements.txt and run the server; docker compose up --build starts the full app. To add it as a skill to a coding agent run npx skills add assafelovic/gpt-researcher. You need an LLM API key and a search API key (for example Tavily).

Pricing and rights

Free under Apache-2.0. Costs come from the LLM and search APIs you use; a standard report costs cents, deep research more. You own the output, but check cited sources before publishing.

Who it is for

Bloggers, analysts and content teams who need researched, sourced drafts or briefs, and developers embedding research in their own apps.

Verdict

One of the most mature open research agents, useful for sourced first drafts. Reports can still misread sources, so treat them as research briefs that need human fact-checking.