story-review-loop

Automated multi-agent code review for pull requests coordinates six specialized agents to review diffs, fix identified blockers, and summarize changes across configurable iterations. The workflow commits fixes locally after each round and pushes a final summary to the PR upon completion.

A multi-round story review loop is an automated git workflow process where six specialized agents iteratively review PR diffs for narrative coherence, technical validation, and script continuity. It runs for a configurable number of rounds, automatically fixing blockers and propagating changes across files before posting a GO/NO-GO summary.

Run automated code review agents on a PR diff by invoking the workflow with a PR number or URL and an iteration count. Six parallel agents focus on distinct quality aspects like propagation, devil's advocate checks, and canonical verification, pushing a final summary after all rounds complete.

Yes, you need a specific PR number or URL and an iteration count to start the story-review-loop. The workflow uses these inputs to fetch PR diffs, coordinate the six review agents across the specified rounds, and push the final summary to the correct pull request.

Yes, you can configure the number of review rounds for your PR workflow by providing an iteration count when invoking the automation. The agents will review, fix, and commit changes locally across the diff for the specified number of rounds before pushing once at the end.

Limitations of automated multi-agent code review include its reliance on orchestrator triggers to initiate rounds and its dependency on a canonical story-review skill for passes. It is specifically designed for story PRs, meaning it may not suit general code reviews outside of narrative or script continuity contexts.