autoresearch

You can automate autonomous research workflows by using a two-loop architecture that alternates between rapid hypothesis experiments and reflective synthesis to scan literature, test hypotheses, and generate paper-level conclusions.

A two-loop research architecture runs an inner loop for rapid experiment iterations and an outer loop for reflective synthesis, continuously updating the research narrative and direction until paper-level conclusions are reached.

You set up autonomous research orchestration by initializing a structured workspace containing a research-state.yaml file and a findings.md file to track progress, then running a mandatory wall-clock continuity loop before experiments begin.

Yes, continuous experiment orchestration operates autonomously by maintaining wall-clock continuity via a loop command or cron heartbeat, ensuring research keeps progressing across time ticks without manual intervention.

Autonomous research routing delegates domain execution like training, evaluation, and data analysis to specialized domain skills while the orchestrator manages state, protocols, and bookkeeping for the overall workflow.

The autonomous loop requires a structured workspace initialization and a continuous wall-clock heartbeat to function; without this mandatory continuity loop, the research orchestration cannot maintain progress or generate paper-level outputs.