auto-review-loop

Automating multi-round research reviews involves orchestrating autonomous cycles that assess work, extract scores, propose fixes, and re-evaluate until acceptance. This skill manages the entire phase-driven workflow, applying iterative assessments until a positive result or maximum round limit is reached.

Yes, autonomous review cycles can be applied to manuscript submissions, code-review pipelines, and experimental reports. The system iteratively evaluates the research work, proposes fixes between rounds, and documents the process to accelerate rigorous evaluation.

Setting up an iterative model evaluation pipeline requires a phase-driven workflow with persistent state, managing review, assessment parsing, fix implementation, and documentation. This skill supports optional human checkpoints during the autonomous cycles to ensure rigorous evaluation standards.

The automated scoring mechanism works by performing round-based assessments to extract scores, proposing targeted fixes based on those results, and re-evaluating the updated work. It continuously loops through these phases until a positive assessment is achieved or the maximum round limit is reached.

The maximum round limit defines the upper boundary for autonomous review cycles to prevent infinite looping. The system continues applying automated scoring, fix proposals, and re-evaluations until either a positive assessment is achieved or this configured MAX_ROUNDS limit is reached.

Yes, the automated review workflow maintains persistent state across all phases. It tracks the transition between Review, Parse Assessment, Implement Fixes, Wait for Results, and Documentation phases to ensure continuous progress throughout the autonomous multi-round review cycles.