Cette page est affichée en anglais. Une traduction française est en cours.
ScriptsMar 28, 2026·2 min de lecture

GPT Researcher — Autonomous Research Report Agent

AI agent that generates detailed research reports from a single query. Searches multiple sources, synthesizes findings, and cites references.

Introduction

GPT Researcher is an autonomous research agent that generates detailed, citation-backed research reports from a single query. It crawls 20+ web sources in parallel, cross-references information to reduce bias, and produces reports exceeding 2,000 words — complete with source citations. Think of it as a research assistant that does hours of googling and reading in minutes.

Works with: GitHub Copilot, Gemini CLI, OpenAI Codex, Ollama ---

Architecture

The system uses a planner + execution agent pattern:

  1. Task agent created based on your research query
  2. Planner generates focused sub-questions for objectivity
  3. Crawler agents gather information per question (parallelized)
  4. Summarizer tracks sources and extracts key findings
  5. Publisher aggregates everything into a structured report

Key Features

  • 📝 Detailed research reports from web and local documents
  • 🖼️ Smart image scraping and AI-generated inline images (via Gemini)
  • 📜 Reports exceeding 2,000 words with citations
  • 🌐 Aggregates 20+ sources for balanced conclusions
  • 🔍 JavaScript-enabled web scraping
  • 📂 Persistent memory and context throughout research
  • 📄 Export to PDF, Word, Markdown
  • 🌳 Deep Research mode with tree-like recursive exploration
  • 🤖 MCP Client integration for specialized data sources

Installation Options

Via pip

pip install gpt-researcher

From source

git clone https://github.com/assafelovic/gpt-researcher.git
cd gpt-researcher
pip install -r requirements.txt
python -m uvicorn main:app --reload
# Visit http://localhost:8000

Docker

docker-compose up --build
# React app at localhost:3000, API at localhost:8000

Usage Examples

Basic research

researcher = GPTResearcher(query="why is Nvidia stock going up?")
await researcher.conduct_research()
report = await researcher.write_report()

Research local documents

export DOC_PATH="./my-docs"
researcher = GPTResearcher(query="analyze these docs", report_source="local")

Supports: PDF, plain text, CSV, Excel, Markdown, PowerPoint, Word.

MCP integration

researcher = GPTResearcher(
    query="top open source research agents",
    mcp_configs=[{
        "name": "github",
        "command": "npx",
        "args": ["-y", "@modelcontextprotocol/server-github"],
        "env": {"GITHUB_TOKEN": os.getenv("GITHUB_TOKEN")}
    }]
)

Deep Research

Tree-like recursive exploration with configurable depth/breadth. Takes ~5 minutes, costs ~$0.40 with o3-mini.


FAQ

Q: What is GPT Researcher? A: AI agent that generates detailed research reports from a single query. Searches multiple sources, synthesizes findings, and cites references.

Q: How do I install GPT Researcher? A: Check the Quick Use section above for step-by-step installation instructions. Most assets can be set up in under 2 minutes.

🙏

Source et remerciements

Created by Assaf Elovic. Licensed under Apache 2.0. gpt-researcher — ⭐ 26,000+ Docs: docs.gptr.dev

Thanks to Assaf Elovic for building an open alternative to deep research tools. Active development with regular updates.

Discussion

Connectez-vous pour rejoindre la discussion.
Aucun commentaire pour l'instant. Soyez le premier à partager votre avis.

Actifs similaires