Automate iterative review-and-fix loops to reduce code quality issues.
npx skills add https://github.com/sato-dev1234/claude-code-sample --skill refine-loop
Please help me install this Agent Skill. Skill: refine-loop Source: https://github.com/sato-dev1234/claude-code-sample/tree/main/skills/refine-loop Command: npx skills add https://github.com/sato-dev1234/claude-code-sample --skill refine-loop
The refine-loop skill automates an iterative review and fix cycle that reduces persistent code quality issues by repeatedly reviewing and applying fixes until no actionable issues remain or the maximum iterations are reached.
Start the refine loop by selecting a type and providing a ticket or file reference, then let the loop run until issues are resolved or the maximum iterations are reached.
High-intent search queries and answers about installing and using this skill.
Automated iterative code review runs a review-and-fix loop that repeatedly applies fixes to code until no actionable issues remain or the maximum iteration limit is reached. It processes code, tests, UI tests, docs, and AC reports.
An iterative review-and-fix loop is an automated workflow that continuously reviews code quality and applies fixes until all actionable issues are resolved. It tracks context like ticket paths, scope, and file references across multiple iterations.
To start the loop, provide a ticket or file reference and define the scope as uncommitted or branch. The workflow then executes guided protocol iterations, saves state per type, and generates a final report upon completion.
Yes, automated iterative review supports multiple artifact types including code, tests, code-tests, UI tests, docs, and AC reports. For documentation generation, you can supply optional knowledge alongside the file path or content.
Yes, the workflow enforces a configurable MAX_ITERATIONS limit to control the review-and-fix cycle. The loop automatically terminates when no actionable issues remain or when it reaches the maximum iteration boundary.
The best way is running an automated multi-type review that executes parallel checks across code, tests, docs, and AC reports. This context-driven approach saves per-type state and applies fixes iteratively until issues are resolved.
yannicklothAutomate iterative review-fix loops until convergence or 12 rounds.
haleyrcSquash merge feature branch commits into main with a Conventional Commit message.
srulytAssesses whether code meets acceptance criteria and passes tests.
alangebRun pyscan, pyanalyze, ruff, and black to generate a unified markdown report.
