Analyze code diffs and produce evidence-based review findings with a verdict.
npx skills add https://github.com/mcevoyinit/agentic-skills --skill reviewer-mcevoyinit
Please help me install this Agent Skill. Skill: reviewer Source: https://github.com/mcevoyinit/agentic-skills/tree/main/skills/coding-modes/reviewer Command: npx skills add https://github.com/mcevoyinit/agentic-skills --skill reviewer-mcevoyinit
Code review is often noisy and can miss real issues. This Skill enforces a read-only posture and delivers concise, evidence-backed findings to help teams focus on meaningful risks.
Activate reviewer mode in your CI or chat interface and provide the code diff or PR for evaluation.
High-intent search queries and answers about installing and using this skill.
Evidence-based code review findings require analyzing the diff alongside test files and related code to identify production, architecture, and security risks. The reviewer skill enforces a read-only posture to surface concrete line-level issues without suggesting edits.
Limiting noise during automated code reviews involves capping output to five evidence-backed findings and classifying risk severity upfront. This signal-over-noise approach ensures developers focus only on critical issues rather than minor stylistic suggestions.
A final verdict for pull request risk assessment should conclude with a single structured decision: APPROVE, REQUEST CHANGES, or NEEDS DISCUSSION. This provides clear guidance on whether the code changes are safe for production deployment.
The code review process does consider test files and system context by reading related configurations and callers. This system-aware approach assesses the true impact of code changes rather than reviewing diffs in isolation.
You can use this to detect security and architecture risks in production code. It analyzes pull request diffs to identify these specific threats, classifying risk types and severity before delivering focused findings.
You should avoid using an automated review mode if you need suggested code edits or fixes. This approach maintains a strictly read-only posture, producing structured findings and verdicts rather than proposing direct modifications to your codebase.
shysssthanhtriAssess pull request changes and summarize risk and impact on execution flows.
s3nex-comCoordinate pull request lifecycle from review through merge with mandatory gates.
leafcoderStandardize Git practices and collaboration workflows across GitHub and Gitee.
act70255Lint and statically analyze Node.js and Python code with ESLint, Pylint, Ruff, MyPy, and Bandit.
