Code Reviewer /code-reviewer

About this skill

Code Reviewer reads your diff like a senior engineer on their third coffee: thorough, specific, and allergic to vague feedback. Every finding gets a severity score, a plain-English explanation, and a proposed fix diff you can apply directly.

It understands context across files — it won't flag a pattern it can see is intentional two files over — and it adapts its strictness to your repo's own conventions, learned from your git history.

Teams wire it into CI so every pull request gets a first-pass review in seconds, freeing human reviewers for architecture and judgment calls. It speaks 14 languages, with the deepest support for Python, TypeScript, Go, and Rust.

Inputs

• diff · string
  Unified diff or file paths to review.
• strictness · string
  lenient | standard | strict. Default: standard.
• focus · array
  Optional focus areas: security, performance, style, tests.

Outputs

• findings · array
  Findings with severity, line refs, explanations.
• fix_diffs · array
  Ready-to-apply unified diffs per finding.
• summary · string
  Overall assessment and risk rating.

Permissions

Deny-by-default: the skill can only use what it declares.

Manifest

{
  "spec": "agentskills.io/v1",
  "name": "code-reviewer",
  "version": "3.0.2",
  "description": "Reviews diffs or files line by line, scores each finding by severity, explains the reasoning, and proposes ready-to-apply fix diffs.",
  "inputs": {
    "diff": {"type": "string", "description": "Unified diff or file paths to review.", "required": true},
    "strictness": {"type": "string", "description": "lenient | standard | strict. Default: standard.", "required": false},
    "focus": {"type": "array", "description": "Optional focus areas: security, performance, style, tests.", "required": false}
  },
  "outputs": {
    "findings": {"type": "array", "description": "Findings with severity, line refs, explanations."},
    "fix_diffs": {"type": "array", "description": "Ready-to-apply unified diffs per finding."},
    "summary": {"type": "string", "description": "Overall assessment and risk rating."}
  },
  "permissions": ["fs:read", "exec:shell"],
  "author": "Patchline Labs",
  "category": "Development"
}

Changelog