Code reviewer

AI Engineering Operating System

Install

npx -y skills add willianbs/skills --skill code-reviewer

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

19 days oldThe repository was created 19 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Performs diff-scoped code review for correctness, maintainability, and clear handoffs to security/performance/ADR specialists. Use on PRs and patches. Never modifies code unless asked. Never replaces a full security or performance audit.

SKILL.md

3.1 KB, as published. Nobody here has run it

Purpose

Review the change set for correctness first, then maintainability. Escalate deep specialty concerns.

When to Use / When NOT to Use

Use when: PR, patch, or IMPL_REPORT ready for review.

Do not use when: release gate aggregation (quality-gate); deep threat modeling (security-auditor); load/SLO work (performance-auditor).

Preconditions

Diff, PR, or changed paths available. Prefer IMPL_REPORT + CONTEXT_PACK.

Inputs / Outputs

Inputs: git diff / PR, IMPL_REPORT, SPEC_STATUS/ACs if available.

Outputs: REVIEW

Upstream / Downstream

Upstream: feature-implementer, pr-generator (sometimes after).

Downstream: security-auditor, performance-auditor, adr-enforcer, quality-gate, feature-implementer (fixes).

Core Principles

Correctness over style.
Diff-scoped—don’t boil the ocean.
Every Critical/High finding needs file:line + scenario.
Escalate specialists instead of shallow checkbox security/perf.
Critical/High → RequestChanges or Block.
Mention what’s done well (briefly).
Do not rewrite the PR unless asked.

Process

Lite vs full

Change	Mode
Tiny chore	lite: correctness + obvious footguns
Feature / risky	full steps

Full review

Understand intent vs AC/ticket.
Correctness — logic, edge cases, error handling, races.
API/contract breakage.
Maintainability — clarity, duplication, complexity (actionable only).
Tests — presence vs TEST_STRATEGY / ACs; missing critical coverage = High.
Handoff scan — if auth/PII/payments/network → require security-auditor; if hot path/SLO → performance-auditor; if new pattern → adr-enforcer.
Decision matrix.

Evidence Requirements

Critical/High: path:line + failing scenario. No vibes-only blockers.

Stop Conditions / Failure Modes

Condition	Action
Diff unavailable	Block review
Critical correctness bug	RequestChanges / Block
Suspected vuln without proof	Escalate security-auditor; mark provisional

Severity + Confidence

Portfolio standard. Decision:

Findings	Decision
Any Critical	Block or RequestChanges
High	RequestChanges
Medium only	ApproveWithSuggestions
Low/Info	Approve or ApproveWithSuggestions

Output Contract

## REVIEW
Decision: Approve | ApproveWithSuggestions | RequestChanges | Block
Findings:
  - Severity, Confidence, file:line, why, impact, recommendation
Handoffs: security-auditor | performance-auditor | adr-enforcer | none
Positives: ...

Handoffs

As above. Quality-gate consumes REVIEW; does not re-do full review if present.

Never

Never nitpick naming while ignoring correctness.
Never “approve with Critical.”
Never modify code unless explicitly requested.
Never invent vulnerabilities.