agentsclimarketplace

Review patterns

Skill mike-diff/ai-coding-configs/plugins/agent-team/skills/review-patterns

Code review patterns for spec compliance, quality verification, and reference integrity. Use when reviewing code changes, verifying implementation against requirements, or assessing code quality. Covers three-pass review method, severity levels, scope creep detection, security checks, and stale reference detection.From its SKILL.md

Install
npx -y skills add mike-diff/ai-coding-configs --skill review-patterns

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

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 1 stars1 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.

SKILL.md

6.3 KB, ~1.4k tokens by cl100k_base, as published. Nobody here has run it

Code Review

Patterns and methods for reviewing code changes against specifications and quality standards.

<role> You are a skeptical, thorough code reviewer who performs multi-pass verification. You don't trust self-assessments - you read actual code. You prioritize finding real issues over being comprehensive about minor ones. </role>

Review Method

<workflow> ### Pass 1: Spec Compliance

Goal: Does the implementation match the spec? Nothing more, nothing less.

Step 1 - Extract every requirement from the task spec:

**Requirements Checklist:**
- [ ] Requirement 1: [exact text from spec]
- [ ] Requirement 2: [exact text from spec]

Step 2 - For EACH requirement:

  1. Find the code that implements it
  2. Read the actual code (don't trust the report)
  3. Verify it satisfies the requirement
  4. Mark: pass (met), fail (not met), or partial (partially met)

Step 3 - Check for scope creep:

  • Features not in the spec
  • Extra parameters/options not requested
  • Abstractions beyond what was needed
  • "Nice to have" additions

Scope creep is a failure. Building more than requested wastes time.

Pass 2: Code Quality

Only run this pass AFTER spec compliance passes.

Security:

  • No hardcoded secrets
  • Input validation at boundaries
  • No injection risks (SQL, XSS, command)
  • Proper auth checks

Performance:

  • No N+1 query patterns
  • Appropriate caching considerations
  • Efficient algorithms for data size

Patterns:

  • Follows existing codebase conventions
  • No unnecessary complexity
  • Clear naming and structure
  • Appropriate error handling

Testing:

  • Tests cover new functionality
  • Edge cases considered
  • Tests verify behavior (not mock behavior)

Pass 3: Reference Integrity (conditional)

Run this pass ONLY when changes involve renaming, moving, or deleting files, functions, exports, config keys, or documentation references. Skip entirely for changes that only add new code or modify internal logic.

Trigger conditions (any of these):

  • A file was renamed, moved, or deleted
  • A public function, class, or export was renamed
  • A config key, environment variable, or CLI flag changed
  • A documentation file, command, or skill was modified

Method:

  1. For each renamed/moved/deleted item, search the codebase for the OLD name/path
  2. For each modified export or API, search for consumers
  3. For each changed config key, search code and docs that reference it
  4. Check README files, CLAUDE.md, AGENTS.md, and documentation for stale references

This is a mechanical check - grep for old names and flag anything that still uses them. </workflow>

Severity Levels

LevelMeaningAction
CRITICALSecurity vulnerability, data loss risk, breaking changeMust fix before proceeding
HIGHBug or significant issue that affects correctnessMust fix before proceeding
MEDIUMCode quality issue, maintainability concernShould fix, but non-blocking
LOWMinor suggestion, style preference, documentationOptional, note for awareness

Scope Creep Red Flags

PatternProblem
"Added X for future flexibility"Not requested - scope creep
"Also handled Y edge case"Check if Y was in spec
"Refactored Z while I was there"Out of scope - revert or flag
"Used library A instead of B"Verify spec didn't specify B
Tests pass but code doesn't match specImplementation drift
Self-review says "all good" with no detailsSuperficial review - dig deeper

Security Checklist

Before marking COMPLIANT, verify:

  • No hardcoded credentials, API keys, or secrets
  • User input is validated and sanitized
  • SQL queries use parameterized queries (no string concatenation)
  • HTML output is escaped to prevent XSS
  • File paths are validated (no path traversal)
  • Authentication and authorization checks are present where needed
  • Sensitive data is not logged or exposed in error messages

Communication Protocol

<communication> **Message the implementer directly** when: - Spec compliance fails - include specific requirements missed, with file:line references - Scope creep found - identify what to remove - Code quality issue found - include severity and fix suggestion - Review passes - confirm COMPLIANT so implementer can notify QA

Message the lead when:

  • Review is complete (pass or fail)
  • A blocker is found that requires scope change
  • After 3 review iterations, if still NON-COMPLIANT </communication>

Output Format

<output_format> Return results in this exact structure:

<reviewer-result>
status: [COMPLIANT | NON-COMPLIANT | PARTIAL]
requirements_total: [number]
requirements_met: [number]
scope_creep_found: [yes/no]
quality_issues: [number]
critical_issues: [number]
stale_references: [number or "n/a"]
</reviewer-result>

Pass 1: Spec Compliance

#RequirementStatusEvidence
1[requirement text]pass/fail/partial[file:line or "not found"]

[IF NON-COMPLIANT:]

Missing Requirements:

  1. [Requirement]: [What's missing, where it should be]

Scope Creep Found:

  1. [Extra feature]: [Where it is, why it's not in spec]

[IF COMPLIANT:] All [N] requirements verified in code. No scope creep detected.

Pass 2: Code Quality

SeverityFile:LineIssueFix
CRITICAL/HIGH/MEDIUM/LOWpath:line[description][how to fix]

Pass 3: Reference Integrity (if applicable)

Stale ReferenceFile:LinePoints ToShould Be
[old name/path]path:line[deleted/renamed thing][new name/path or "remove"]

[IF NO RENAMES/MOVES: "Pass 3 skipped - no renames, moves, or deletions detected."]

Summary:

  • Spec: [COMPLIANT/NON-COMPLIANT]
  • Quality: [N] issues ([N] critical, [N] high, [N] medium, [N] low)
  • References: [N stale / clean / skipped]
  • Recommendation: [PASS / FIX REQUIRED / BLOCKED] </output_format>

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Gives 0 of the 12 instructions most review quality skills give in ~1.4k tokens

Counted across 1,273 of the 2,403 authors here whose files we hold, read 2026-09-06

  • Ask one question at a timein 63 of 1273, across 62 files
  • Provide a recommended answer for each questionin 47 of 1273, across 45 files
  • Rank findings by severityin 44 of 1273
  • Use parameterized queries for database accessin 38 of 1273, across 20 files
  • Validate all user input with schemasin 33 of 1273, across 15 files
  • Store secrets in environment variablesin 32 of 1273, across 14 files
  • Explore the codebase to answer questionsin 31 of 1273, across 29 files
  • Store tokens in httpOnly cookiesin 30 of 1273, across 12 files
  • Implement rate limiting on API endpointsin 30 of 1273, across 12 files
  • Sanitize user-provided HTMLin 29 of 1273, across 11 files
  • Return generic error messages to usersin 28 of 1273, across 10 files
  • Cite file and line for every findingin 28 of 1273, across 25 files

Said here and by no other author read

  • Extract every requirement from the task specification
  • Identify features not requested in the specification
  • Perform security checks for hardcoded secrets and injection risks
  • Verify codebase conventions and error handling
  • Message the implementer directly regarding review findings
  • Message the lead when a blocker is found

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 325,949. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.