skills.rest
ZaxbyHub
ZaxbyHubOFFICIAL
Explore all skills by this author →
GitHub

reviewing-code-core

Reviews code changes for correctness bugs, AI-generated smells, and unsupported claims with file-line evidence.

Updated Mar 30, 2026
Search Tags:#quality assurance#code review#validation#bug detection#evidence-based#ai code
One-click install
npx skills add https://github.com/ZaxbyHub/ragappv3 --skill reviewing-code-core-zaxbyhub
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: reviewing-code-core
Source: https://github.com/ZaxbyHub/ragappv3/tree/main/.claude/skills/reviewing-code-core
Command: npx skills add https://github.com/ZaxbyHub/ragappv3 --skill reviewing-code-core-zaxbyhub

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Code reviews often miss hidden bugs, unwired functionality, and AI-generated code that looks plausible but fails in production. This Skill enforces an evidence-first review process that validates every claim against actual code before approving changes. ## Core Features & Use Cases - Evidence-Based Review: Requires exact file and line citations for every defect found, eliminating speculative feedback. - AI-Slop Detection: Identifies hallucinated dependencies, fake abstractions, placeholder code, and happy-path-only logic common in LLM-generated code. - Claim Validation: Cross-checks release notes, docs, and comments against actual implementation to catch mismatches. - Use Case: When reviewing a PR that adds a new API endpoint, this Skill verifies the route is wired, the response model matches the SQL query, frontend types align with backend serialization, and no placeholder logic remains. ## Quick Start Review the current diff for correctness bugs, unwired functionality, and unsupported claims, then provide a verdict with file-line evidence.

Frequently Asked Questions about reviewing-code-core

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I review code for AI-generated bugs?▼

Use evidence-first review by checking for hallucinated dependencies, fake abstractions, placeholder code, and happy-path-only logic. Verify every claim in comments and docs against actual implementation with exact file-line citations.

What is evidence-first code review?▼

Evidence-first review requires exact file and line citations for every defect found. It treats code as plausible until verified, never approves without stating what was checked, and validates all claims against actual implementation.

How to detect unwired functionality in code?▼

Check that routes, commands, exports, schemas, and handlers are actually connected to the application. Look for placeholder functions, TODO comments, dead abstractions, and tests that assert existence instead of behavior.

Does this work for reviewing database-backed API changes?▼

Yes, it verifies response model fields match SQL SELECT statements, checks frontend API types against backend serialization, and validates that renamed routes or models don't break consumer coupling.

What are AI code smells to look for?▼

Common AI smells include mapping-hallucination, phantom dependencies, stale API usage, context rot against local conventions, unnecessary async patterns, and generated duplication without understanding.

Related Skills

Explore More Software Engineering Skills→
kartuli-appkartuli-app

pr-title-and-description

Generate conventional-commit PR titles and markdown descriptions from branch diffs.

Official
Intermediate
AlmogMaayanAlmogMaayan

ptp-receiving-code-review

Validates code review findings against current repository content before applying fixes.

Community
Intermediate
LynxLabVNLynxLabVN

requesting-code-review

Scan code for security concerns and quality issues before committing.

Official
Advanced
flonatflonat
127

verify-math

Verify mathematical claims through numerical, symbolic, and formal Lean proofs.

Community
Advanced