Review implementation diffs across correctness, readability, architecture, security, and performance.
npx skills add https://github.com/leing2021/super-pi --skill 04-review
Please help me install this Agent Skill. Skill: 04-review Source: https://github.com/leing2021/super-pi/tree/main/skills/04-review Command: npx skills add https://github.com/leing2021/super-pi --skill 04-review
This Skill helps you catch bugs, design issues, and risky changes by reviewing code evidence against the implementation diff, the prior plan artifacts, and existing solution patterns before you commit.
review_router.docs/solutions/ for related solution evidence.agent-browser) for visual/functional/UX issues and optionally add regression tests.Use the 04-review skill after you finish implementing changes and want a structured, evidence-first review before committing.
High-intent search queries and answers about installing and using this skill.
To run an evidence-first code review, you provide your implementation diff and optional plan artifacts. The review loads layered rules, searches solution patterns, and emits structured findings with severity, evidence, and recommended actions.
An evidence-first code review checks your diff across five axes: correctness, readability, architecture, security, and performance. It selects reviewer personas based on change scope and diff metadata to produce targeted findings instead of vague opinions.
Yes, an autofix-and-re-review loop applies autofixable findings and re-reviews the changed lines. This automated review loop supports up to three iterations to ensure your code meets the required standards before final decisions.
Yes, after the initial code review, the workflow can continue into a browser QA mode. This mode uses agent-browser to identify visual, functional, and UX issues, and can optionally add regression tests for your implementation.
No, plan artifacts are optional but enhance the review's context awareness. The skill loads relevant plan artifacts when present and searches existing solution directories for related evidence to improve finding accuracy.
lukasdeveloperaccReview code diffs and implementation evidence in Korean via Codex.
bacocoExecute Playwright/agent-browser YAML visual tests and detect UI regressions.
sommioOrchestrate multi-agent code reviews and merge findings into a single report.
randoneeringClassify dbt-core to Fusion migration errors into categories A-D.