loop-review-skill-until-fixed-point

Automating iterative peer reviews requires running isolated review passes against a target document until all reviewers report NO ISSUES. This skill loops through context-isolated iterations, recording per-iteration outcomes to enforce convergence on policy papers, technical specifications, or code reviews.

A fixed point in automated document review is reached when all reviewers declare NO ISSUES after repeated passes. This skill enforces this convergence by applying context-isolated iterations to a target document, ensuring collaborative consensus before finalizing.

Running repeated reviews on a SKILL.md file requires providing the target path and a max_iterations cap. The skill executes isolated iterations of the review command, automatically stopping when all reviewers return no issues or the iteration limit is met.

Yes, automated review loops support collaborative review workflows across technical specifications, code reviews, and policy papers. The skill applies isolated review passes to these documents, repeating until all reviewers converge on no issues.

Reviews fail to converge when findings leak between iterations or the max_iterations cap is too low. This skill prevents context leakage by isolating each review pass, ensuring prior findings do not influence subsequent iterations, and records per-iteration outcomes for tracking.

Yes, a max_iterations cap is required to bound the automated review loop. The skill uses this cap alongside a target SKILL.md path to execute isolated review iterations, preventing infinite loops while converging on a no-issues fixed point.