What problem does it solve? Autonomous research pipelines risk promoting methods based on weak or screening-only evidence. This Skill enforces a final promotion gate that audits full-budget confirmation results before any method is frozen or any performance claim is written. ## Core Features & Use Cases - Evidence Audit: Runs an evidence-audit against the baseline, frozen candidate, full results, and proposed claim, verifying that held-out evidence was not used for final selection. - Bounded Claim Writing: Produces a result-to-claim table with the exact supported claim, unsupported extensions, missing evidence, and revised bounded wording. - Decision Routing: Applies audit outcomes to route the workflow — freeze on audit_pass, return to confirmation on audit_fix, or fall back to mapping/literature on audit_fail. - Use Case: After a candidate method passes full-budget confirmation, use this Skill to verify the evidence, record a reproducibility receipt (commit, config, seeds, environment), and decide whether to freeze the method and hand off to paper writing. ## Quick Start Run the promotion review on the confirmed candidate to audit its evidence and decide whether to freeze the method or route it back for correction.
