What problem does it solve? Iterative research and engineering work often degrades into ad hoc shell loops, hand-kept metric logs, and lost experiment history. This Skill enforces a managed, gated workflow where every candidate is committed, evaluated through a pinned TORC profile, and archived as durable evidence, so results stay comparable and reproducible. ## Core Features & Use Cases - Managed study lifecycle: Creates NestedText study contracts with objective, acceptance rules, and budgets, then drives candidates through waterology study create, enqueue, advance, and watch. - Durable evidence and reproducibility: Seals run archives, exports RO-Crates, and verifies final deliverables with deliverable register and reproduce in an isolated directory. - Engineering and research modes: Supports both specification-driven optimization and scientific question answering, with an optional experiment-tree discipline for multi-decision studies. - Use Case: A researcher wants to calibrate a hydrological model against a benchmark. The Skill registers the evaluation workflow, saves a contract with a metric target and iteration budget, then iterates committed candidate branches until acceptance or budget limits are reached. ## Quick Start Ask the assistant to start an autoresearch study that optimizes your model's benchmark metric with a defined iteration budget and let it set up the managed workflow.
