autoresearch

To manage autonomous AI research projects, use a two-loop architecture that handles the full research lifecycle. It orchestrates rapid experiment iterations, synthesizes results, and steers research direction from initial literature search to final paper publication.

The two-loop architecture is a management mechanism for autonomous research. One loop drives rapid experiment iterations and execution, while the second loop synthesizes results and identifies patterns to continuously steer the research direction.

Set up your workspace by defining a research question and selecting initial hypotheses. The system then initiates the project, routing domain-specific skill execution and tracking experiments to support continuous agent operation.

Yes, autonomous research synthesis requires external search tools. The system depends on integrations like exa_mcp, arxiv, semanticscholar, and crossref to conduct literature surveys and gather necessary research data for hypothesis formation.

Yes, the system supports continuous agent operation for multi-hypothesis research efforts. It routes tasks to domain-specific skills for execution, allowing agents to run continuously while tracking and synthesizing multiple experiments.

The two-loop architecture requires predefined workspace setup and external dependencies for literature search. It is limited by the need for continuous agent operation and relies on domain-specific skill routing to execute complex experiments.