Skip to content
D-CSIL

Research Domain

Autonomous Research Systems

The Autonomous R&D Lab: objective → plan → agents → experiments → evidence → critique → revision, with claim gates and immutable audit logs.

EXPERIMENTALCURRENT EVIDENCE LEVEL

Research Question

Can a governed loop of specialist agents run real experiments and only publish claims that survive skeptical review?

The lab engine assigns nine specialist roles with scoped permissions, sandboxes experiments, and forces every claim through Skeptic, Statistical Evaluator, and Auditor review gates before it counts as a result.

The MVP is feature-complete and self-validates with a toy experiment playbook; live-LLM quality remains the honest gap — the default provider is still a mock.