skills.rest
Thigh-Food
Thigh-FoodOFFICIAL
Explore all skills by this author →
GitHub

adversarial-reviewer

Reviews code adversarially to find bugs across logic, security, concurrency, and data integrity categories.

Updated Sep 8, 2026
Search Tags:#code review#pull request#concurrency#security audit#bug hunting#adversarial review#severity triage
One-click install
npx skills add https://github.com/Thigh-Food/emdash --skill adversarial-reviewer-thigh-food
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: adversarial-reviewer
Source: https://github.com/Thigh-Food/emdash/tree/main/skills/adversarial-reviewer
Command: npx skills add https://github.com/Thigh-Food/emdash --skill adversarial-reviewer-thigh-food

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Standard code reviews often miss bugs because reviewers give code the benefit of the doubt. This Skill forces a hostile review posture that assumes bugs exist and hunts for them with concrete triggers, producing actionable findings instead of vague suggestions. ## Core Features & Use Cases - Structured Bug Hunt: Works through seven categories in order — logic errors, edge cases, error handling, state and concurrency, security, data integrity, and resource management. - Severity-Ranked Output: Reports each finding with file and line, category, severity (CRITICAL to LOW), a concrete trigger scenario, and a minimal fix. - Use Case: Before merging a risky PR touching payment logic, ask for an adversarial review to surface race conditions, missing validation, and swallowed errors that a friendly review would overlook. ## Quick Start Review this pull request adversarially and find every bug with concrete triggers and severity ratings.

Frequently Asked Questions about adversarial-reviewer

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

FAQPage Schema
How do I get a code review that actually finds bugs?▼

Use an adversarial review approach that assumes the code is broken and requires concrete proof for each finding. This Skill works through seven bug categories and reports each issue with a trigger scenario, severity rating, and minimal fix.

What kinds of bugs does adversarial code review catch?▼

It covers logic errors, edge cases, error handling failures, race conditions and stale closures, security issues like injection and missing authorization, data integrity problems, and resource leaks such as missing cleanup or unbounded growth.

How are findings from the review reported?▼

Each bug is reported with a title, file and line number, category, and severity from CRITICAL to LOW. Every finding includes a concrete trigger scenario and a minimal suggested fix, ordered by severity.

When should I not use adversarial code review?▼

Skip it when you want style feedback, naming suggestions, feature ideas, or refactoring advice. This review deliberately excludes style and design commentary and only reports demonstrable bugs.

What happens if the review finds no bugs?▼

The reviewer reports "No bugs found" and stops. It does not manufacture issues to appear thorough, so silence on any code section constitutes approval of that section.

Related Skills

Explore More Software Engineering Skills→
wizelinewizeline

code-review-orchestrating

Orchestrate code reviews across security, performance, correctness, and maintainability.

Official
Advanced
EotelEotel

code-review

Scan uncommitted code changes for vulnerabilities and quality defects before commit.

Community
Advanced
phraphra

code-review

Review modified code files and auto-fix mechanical issues.

Community
Advanced
CagesThrottleUsCagesThrottleUs

property-based-testing

Generates property-based and fuzz tests for functions with checkable invariants across five language ecosystems.

Community
Intermediate