It is a Claude Code skill that automates the process of running code reviews and applying fixes in a continuous loop until the code meets your specified quality standards.

What modes does Candid Loop support?

It offers 'Auto' for fast automated fixes, 'Review-Each' for granular manual approval of every change, and 'Interactive' for full control over skipping or ignoring specific issues.

Can I exclude specific issues from the loop?

Yes, you can ignore issues by specific ID, regex title patterns, or entire categories via CLI flags or the .candid/config.json file.

Does it work with existing team configurations?

Yes, it respects both project-level and user-level settings in .candid/config.json, allowing teams to standardize their review categories and ignore lists.

What happens if the loop gets stuck in a cycle?

The skill includes a max-iterations setting (defaulting to 5) to prevent infinite loops and provides a detailed summary of any remaining issues if the limit is reached.

Candid Loop - Automated Review Cycle

Name: Candid Loop - Automated Review Cycle
Author: ron-myers

byron-myers

•

安全与测试

Automates the iterative cycle of code review and bug fixing to systematically reach a clean, issue-free code state.

Candid Loop is a powerful automation skill for Claude Code that streamlines the software quality assurance process by running the candid-review tool in a continuous loop. It automates the tedious fix-review-fix cycle, allowing developers to either automatically apply suggested fixes or interactively manage issues until the codebase is clean or a maximum iteration count is reached. With support for persistent ignore lists, category-based enforcement, and deep configuration through JSON, it provides a robust framework for maintaining high architectural standards and resolving everything from critical security holes to minor code smells.

主要功能

01Three execution modes: Auto, Review-Each, and Interactive for granular control

02Configurable iteration limits to prevent infinite loops while ensuring code quality

0314 GitHub stars

04Persistent ignore lists that support issue IDs, regex title patterns, and categories

05Granular category filtering (Critical, Major, Standards, Smell, Edge Case, Architectural)

06Deep integration with project-level and user-level .candid/config.json settings

使用场景

01Automating the cleanup of legacy codebases by resolving all 'Smell' and 'Standards' issues in bulk

02Performing interactive architectural reviews where specific patterns can be ignored or batch-processed

03Enforcing strict 'Critical' and 'Major' issue resolution during CI/CD or pre-release phases

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add ron-myers/candid candid-loop

For use in Claude.ai and ChatGPT

Download Skill