gpt-researcher

Automated web research generates cited reports by using an LLM agent with a planner-executor-publisher pattern to produce structured Markdown. You instantiate a research agent with a query, execute the research task, and write the final cited report.

The planner-executor-publisher pattern is an autonomous architecture where the agent plans research strategy, executes web and local data retrieval, and publishes a comprehensive cited report. This structure ensures fast, structured, and less biased information gathering.

Yes, autonomous research supports local documents and MCP data sources alongside web sources. The agent integrates these retrievers and embeddings to enrich the final research report, enabling comprehensive similarity discovery across mixed data environments.

Yes, autonomous web research requires a Python runtime with the specific research package installed. You also need access to an LLM provider, configured retrievers, valid API keys, and network access to gather information and generate cited reports.

You integrate a research agent API by installing the package, creating a research instance with a task query, and calling the asynchronous research execution and report writing methods. This returns a Markdown report you can embed into your products.