What is multi-agent-research-assistant-?

This is a Python project that automates the production of research reports by running a chain of AI agents. You give it a topic, and it searches the web, checks facts, and writes a structured report without any further input from you. The pipeline has four roles. A Researcher agent takes the topic, breaks it into sub-questions, and searches the web for raw information. A Fact-Checker agent goes through each claim from the Researcher, validates it against sources, and removes anything unverified or contradictory. A Writer agent takes the verified facts and structures them into a formatted report with an introduction, body sections, and citations. An Orchestrator manages the whole process: passing data between agents, handling failures, and logging how long each step takes. Each agent is built using the Anthropic Claude API. They share data through a common structure called ResearchContext rather than through global variables. You run the tool from the command line by providing a topic and optional flags for depth (quick, standard, or deep), output format (Markdown, PDF, or JSON), and which Claude model to use. The generated report is saved to a folder on your computer. You need an Anthropic API key to use the tool. The key goes into a configuration file. The README includes a quick setup walkthrough covering how to create a Python virtual environment, install dependencies, and add the key. Items on the roadmap that are not yet complete include PDF export, parallel research on multiple subtopics at once, a web interface, and integrations with tools like Slack or Notion. The project is MIT-licensed.