Code Review Checklist
shareAI-lab/learn-claude-code
Reviews code against a five-part checklist covering security, correctness, performance, maintainability and testing, and reports findings in a fixed format.
A skill your agent uses to conduct aggressive, adversarial code reviews on git changes, branches, or PRs to detect critical edge cases, race conditions, security holes, and data loss risks.
Install Codex Adversarial Review
$ npx skills add WrongStack/WrongStack --skill codex-adversarial-review -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install WrongStack/WrongStack codex-adversarial-review --agent claude-codeProject scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).
$ git clone --depth 1 https://github.com/WrongStack/WrongStack.git skills-src && mkdir -p .claude/skills && cp -r skills-src/packages/core/skills/codex-adversarial-review .claude/skills/codex-adversarial-review && rm -rf skills-srcUse ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.
Claude Code skills documentation · loads skills from .claude/skills/
Install the "codex-adversarial-review" agent skill from https://github.com/WrongStack/WrongStack/tree/main/packages/core/skills/codex-adversarial-review into .claude/skills/codex-adversarial-review/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "codex-adversarial-review", then confirm the skill loads.Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$skill-installer install https://github.com/WrongStack/WrongStack/tree/main/packages/core/skills/codex-adversarial-reviewType this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.
$ npx skills add WrongStack/WrongStack --skill codex-adversarial-review -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install WrongStack/WrongStack codex-adversarial-review --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/WrongStack/WrongStack.git skills-src && mkdir -p .agents/skills && cp -r skills-src/packages/core/skills/codex-adversarial-review .agents/skills/codex-adversarial-review && rm -rf skills-srcUse ~/.agents/skills/ instead of .agents/skills for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "codex-adversarial-review" agent skill from https://github.com/WrongStack/WrongStack/tree/main/packages/core/skills/codex-adversarial-review into .agents/skills/codex-adversarial-review/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "codex-adversarial-review", then confirm the skill loads.Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add WrongStack/WrongStack --skill codex-adversarial-review -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install WrongStack/WrongStack codex-adversarial-review --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/WrongStack/WrongStack.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/packages/core/skills/codex-adversarial-review .cursor/skills/codex-adversarial-review && rm -rf skills-srcUse ~/.cursor/skills/ instead of .cursor/skills for a personal install.
Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/
Install the "codex-adversarial-review" agent skill from https://github.com/WrongStack/WrongStack/tree/main/packages/core/skills/codex-adversarial-review into .cursor/skills/codex-adversarial-review/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "codex-adversarial-review", then confirm the skill loads.Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gemini skills install https://github.com/WrongStack/WrongStack.git --path packages/core/skills/codex-adversarial-review--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.
$ npx skills add WrongStack/WrongStack --skill codex-adversarial-review -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install WrongStack/WrongStack codex-adversarial-review --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/WrongStack/WrongStack.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/packages/core/skills/codex-adversarial-review .gemini/skills/codex-adversarial-review && rm -rf skills-srcUse ~/.gemini/skills/ instead of .gemini/skills for a personal install, then run /skills reload.
Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/
Install the "codex-adversarial-review" agent skill from https://github.com/WrongStack/WrongStack/tree/main/packages/core/skills/codex-adversarial-review into .gemini/skills/codex-adversarial-review/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "codex-adversarial-review", then confirm the skill loads.Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gh skill install WrongStack/WrongStack codex-adversarial-reviewInstalls for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).
$ npx skills add WrongStack/WrongStack --skill codex-adversarial-review -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
$ git clone --depth 1 https://github.com/WrongStack/WrongStack.git skills-src && mkdir -p .github/skills && cp -r skills-src/packages/core/skills/codex-adversarial-review .github/skills/codex-adversarial-review && rm -rf skills-srcUse ~/.copilot/skills/ instead of .github/skills for a personal install. Commit .github/skills so cloud agent and code review can use it.
GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/
Install the "codex-adversarial-review" agent skill from https://github.com/WrongStack/WrongStack/tree/main/packages/core/skills/codex-adversarial-review into .github/skills/codex-adversarial-review/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "codex-adversarial-review", then confirm the skill loads.GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add WrongStack/WrongStack --skill codex-adversarial-review -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install WrongStack/WrongStack codex-adversarial-review --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/WrongStack/WrongStack.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/packages/core/skills/codex-adversarial-review .opencode/skills/codex-adversarial-review && rm -rf skills-srcUse ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.
OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/
Install the "codex-adversarial-review" agent skill from https://github.com/WrongStack/WrongStack/tree/main/packages/core/skills/codex-adversarial-review into .opencode/skills/codex-adversarial-review/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "codex-adversarial-review", then confirm the skill loads.OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
Facts
- Skill name
codex-adversarial-review- GitHub stars
- 371
- Token cost
- ~2k tokens
- SKILL.md length
- 657 words
- Files
- 1
- Skills in repo
- 100
- Repo updated
- First seen
- Licence
- MIT
At a glance
A skill your agent uses to conduct aggressive, adversarial code reviews on git changes, branches, or PRs to detect critical edge cases, race conditions, security holes, and data loss risks.
- Works in 2 steps: Adversarial Audit Finding Template → Idempotency & Webhook Double-Spend…
- Conduct aggressive
- SKILL.md covers Selection card, Overview, Rules and Pre-Flight Verification Runbook, plus 6 more sections
- Calls git
What it does
Codex Adversarial Review is an agent skill from WrongStack/WrongStack. Use this skill to conduct aggressive, adversarial code reviews on git changes, branches, or PRs to detect critical edge cases, race conditions, security holes, and data loss risks. Triggers: user mentions "adversarial review", "break the code", "find vulnerabilities", "security audit PR", "race condition check", "ship gate", "pre-merge audit".
Its SKILL.md is about 2k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.
It sits in Development, covering Async programming, Code review and Security review. It works with Git. The repository describes itself as: An AI coding agent that reads your code, edits files, runs commands, and reasons through bugs — across a terminal REPL, a full-screen TUI, and a browser UI, while you keep your… The licence is MIT.
When your agent uses it
- Conduct aggressive
- Adversarial code reviews on git changes
- PRs to detect critical edge cases
- Race conditions
Example prompts
- “adversarial review”
- “break the code”
- “find vulnerabilities”
- “/codex-adversarial-review”
Workflow steps
2 steps, taken from the step headings in SKILL.md.
What it can do on your machine
Read from SKILL.md and the folder at commit a744bdc. It shows what the files ask for, not the result of running them.
Tool permissions
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
Runs code
Shell commands in SKILL.md call:
git
From the folder's file list and the shell code blocks in SKILL.md.
Network
No URLs in SKILL.md. Its commands use git, which can reach the network depending on how they are called.
From URLs in SKILL.md, links to its own repository left out.
Credentials
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.
Context cost
Codex Adversarial Review loads about 2k tokens when it runs. Until then it costs about 93 tokens; SKILL.md has 657 words of instructions outside code blocks.
Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.
Safety
Auto-check passed
The automated check found no risky patterns in SKILL.md.
Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.
SKILL.md
The full file from WrongStack/WrongStack at commit a744bdc, republished under its MIT licence (© WrongStack). 657 words, ~1,958 tokens.
Download SKILL.mdSave it as
.claude/skills/codex-adversarial-review/SKILL.md (or your agent's skills folder).- name
- codex-adversarial-review
- description
- Use this skill to conduct aggressive, adversarial code reviews on git changes, branches, or PRs to detect critical edge cases, race conditions, security holes, and data loss risks. Triggers: user mentions "adversarial review", "break the code", "find vulnerabilities", "security audit PR", "race condition check", "ship gate", "pre-merge audit".
- version
- 1.2.1
- required-capabilities
- filesystem.read
- optional-capabilities
- version-control.manage
- trigger
- Use this skill to conduct aggressive, adversarial code reviews on git changes, branches, or PRs to detect critical edge cases, race conditions, security…
- metadata.routing-group
- quality
Codex Adversarial Code Review Architecture
Selection card
- Task: Review adversarially and propose approval-gated fixes. / TR: Karşıt bakışla incele ve onay gerektiren düzeltme öner.
- Start: Identify the scope and obtain an executable before-proof or review evidence.
- Finish: apply the acceptance checks below; report observed results and unresolved constraints.
Overview
The purpose of an adversarial code review is not to validate the author's work, but to aggressively break confidence in the change by uncovering subtle, high-cost, or user-visible failure modes before the code reaches production. Reviewers adopt the mindset of an attacker, a chaotic network environment, and an overloaded database.
Rules
- Pre-flight: Inspect repo diff & live security advisories first. Read the full git diff, surrounding call sites, and recent dependency changes in
package.json/ lockfiles. Cross-reference suspicious patterns or third-party packages against live security advisories (e.g. GitHub Advisory Database, CVE feeds) before auditing logic. - Default to skepticism. Assume the change fails in high-risk ways until concrete evidence proves safety.
- Focus on high-damage attack surfaces: auth boundaries, tenant isolation, data loss/corruption, idempotency gaps, and race conditions.
- Ignore superficial style, formatting, and naming feedback; report only material, high-impact defects.
- Strict Stop Rule: After presenting review findings, STOP. Never modify code or auto-apply fixes without explicit confirmation from the user.
- Issue a clear ship decision:
BLOCKING_RISKS_DETECTEDif blocking risks exist, orAPPROVEonly if no substantive flaws can be proven. - Construct reproducible proof-of-concept payloads or call sequences for every flagged vulnerability.
- Verify database mutations are wrapped inside atomic transactions with rollbacks on failure.
Pre-Flight Verification Runbook
Before beginning the adversarial audit:
- Extract Changed Files and Git Diffs:
bash git diff --name-only origin/main...HEAD git diff origin/main...HEAD - Scan for New or Modified Third-Party Dependencies:
bash git diff origin/main...HEAD -- package.json pnpm-lock.yaml Cargo.toml - Check for Security Advisories:
- Query npm audit or vulnerability registries for any newly introduced packages.
The 5 Critical Attack Vectors
| Attack Vector | What to Look For | Real-World Impact |
|---|---|---|
| 1. Broken Object-Level Authorization (IDOR) | Fetching records by ID (/api/items/:id) without validating tenant / user ownership | Data leakage across tenant boundaries |
| 2. Race Conditions & Concurrency Gaps | Read-modify-write patterns without database-level row locks or atomic increments | Account double-spending, inventory overselling |
| 3. Idempotency Gaps | Webhooks (e.g. Stripe, billing) that process charges without unique event ID deduplication | Duplicate charges, duplicated user credits |
| 4. Partial Failure & Missing Rollbacks | Mutating DB table A, calling external service B, then updating table C without a transaction | Inconsistent system state and silent data corruption |
| 5. Resource Exhaustion / Unbounded I/O | Missing LIMIT on queries, unescaped Regex on user strings (ReDoS), unbounded file uploads | Server CPU freeze, memory crash, denial of service |
Show full SKILL.md (239 more words)Show less
Patterns
1. Adversarial Audit Finding Template
### [CRITICAL] Race Condition in Quota Deduction
- **File**: src/billing/quota.ts:45-62
- **Attack Vector**: Concurrent HTTP requests bypass usage limits
- **Proof of Concept**:
Sending 10 concurrent requests when `remainingCredits = 1`:
1. Request A reads remainingCredits = 1
2. Request B reads remainingCredits = 1
3. Request A executes operation and sets remainingCredits = 0
4. Request B executes operation and sets remainingCredits = 0
Result: User executes 2 operations while only paying for 1.
- **Remediation**:
Replace application-level check with an atomic database query:
`UPDATE users SET credits = credits - 1 WHERE id = $1 AND credits >= 1 RETURNING credits;`2. Idempotency & Webhook Double-Spend Vulnerability
// ❌ VULNERABLE: Stripe webhook without idempotency or atomic transactions
export async function handleInvoicePaid(event: StripeEvent) {
const invoice = event.data.object;
const user = await db.query.users.findFirst({ where: eq(users.customerId, invoice.customer) });
// Vulnerability: If Stripe sends duplicate webhooks, credits are added twice!
await db.update(users).set({ credits: user.credits + 100 }).where(eq(users.id, user.id));
}
// ✅ SECURE & RESILIENT: Idempotency table + atomic transaction
export async function handleInvoicePaidSecure(event: StripeEvent) {
const invoice = event.data.object;
await db.transaction(async (tx) => {
// 1. Guard against duplicate webhook deliveries
const processed = await tx.insert(processedEvents).values({ id: event.id }).onConflictDoNothing();
if (!processed.rowCount) return; // Already handled, exit safely
// 2. Atomic credit increment
await tx.update(users)
.set({ credits: sql`credits + 100` })
.where(eq(users.customerId, invoice.customer));
});
}Anti-patterns
- Auto-applying fixes without confirmation: The reviewer's role is to stress-test and reveal failure modes, not to silently alter application code.
- Reporting stylistic nits: Commenting on indentation, variable naming, or comment phrasing dilutes focus from actual exploits and critical bugs.
- Assuming input is sanitized: Never assume client-provided IDs, numbers, or JSON payloads are validated upstream.
- Assuming third-party SDK calls succeed: External network requests can time out, return 500s, or drop connections midway through operations.
Before returning
- Full diff analyzed against main branch
- Every finding cites exact file and line ranges
- Concrete failure triggers and real-world consequences detailed
- Explicit ship decision rendered (
BLOCKING_RISKS_DETECTEDorAPPROVE) - Strict Stop Rule respected: No code edits executed without user confirmation
Review boundary and decision
Pin the exact change set and identify the implementation's actual trust/resource boundaries. Review owned source and bounded defensive checks; do not construct offensive workflows or contact live third-party targets. Every material finding needs a reachable condition, impact and confidence level; missing evidence is a validation gap. An approve verdict means no confirmed blockers in the examined scope, not a proof that every failure mode is impossible. The stop rule remains: return findings and wait for explicit confirmation before applying fixes.
Skills in scope
security-scanner— for running static AST checks and credential pattern scanstesting— for writing regression tests proving discovered vulnerabilitiestech-stack— for auditing supply chain packages and dependencies
© WrongStack, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file
Files
Just SKILL.md in packages/core/skills/codex-adversarial-review of WrongStack/WrongStack.
Open the folder on GitHubat commit a744bdc
Compare with similar skills
Codex Adversarial Review next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.
| Skill | Stars | Used in | Tokens | Auto-check | Licence | Repo updated |
|---|---|---|---|---|---|---|
| Codex Adversarial Review this skillWrongStack/WrongStack | 371 | — | ~2k | Automated safety check: Pass | MIT | |
| Code Review ChecklistshareAI-lab/learn-claude-code | 78k | 4 repos | ~1.1k | Automated safety check: Pass | MIT | |
| Requesting Code ReviewHezaoHezao/poirot | 249 | 5 repos | ~1.6k | Automated safety check: Pass | MIT | |
| Openqodexopenqodex/openqodex | 524 | — | ~2.5k | Automated safety check: Pass | Apache-2.0 | |
| Code Review with Beads Tasksmaslennikov-ig/claude-code-orchestrator-kit | 260 | — | ~2k | Automated safety check: Pass | Custom licence | |
| Review Codetobihagemann/turbo | 408 | — | ~3.3k | Automated safety check: Pass | MIT |
Similar skills
Requesting Code Review
HezaoHezao/poirot
Pre-commit review: security scan, quality gates, auto-fix. An agent skill from HezaoHezao/poirot.
Openqodex
openqodex/openqodex
Code review for the current change, before it is pushed. An agent skill from openqodex/openqodex.
Code Review with Beads Tasks
maslennikov-ig/claude-code-orchestrator-kit
Reviews staged changes, a branch, a PR or a path for bugs, security gaps and performance issues, then writes an evidence-based report and creates Beads tasks.
Review Code
tobihagemann/turbo
Review code for bugs, security vulnerabilities, API misuse, consistency issues, simplicity problems, or test coverage gaps and low-value tests by running internal reviews and a peer review in…
Review
softspark/ai-toolkit
Reviews code for quality, security, correctness. An agent skill from softspark/ai-toolkit.
More from WrongStack/WrongStack
All 100 skills in this repoTech Stack
WrongStack/WrongStack
Validate and upgrade dependencies against live registries and official migration guides in any ecosystem.
Skill Creator
WrongStack/WrongStack
Create, improve and validate WrongStack SKILL.md bundles with precise discovery, progressive resources and current runtime contracts.
Bug Hunter
WrongStack/WrongStack
A skill your agent uses when scanning source code for bugs, anti-patterns, code smells, or quality issues in a codebase, or when running a proof-driven bug hunt that must find, prove, fix, and…
Design Craft
WrongStack/WrongStack
Design or substantially improve user-facing interfaces with a product-specific visual direction, content hierarchy, and rendered critique.
Design Critique
WrongStack/WrongStack
A skill your agent uses to audit an interface that already exists and say precisely why it looks generated, templated, or unfinished — a scored rubric across composition, typography, color, states…
Mailbox Bridge
WrongStack/WrongStack
A skill your agent uses when external coding agents (Claude Code, Aider, custom scripts) need to participate in the project's shared WrongStack mailbox, or when a user asks to "expose the mailbox"…
Related
Works with
Categories
Questions about Codex Adversarial Review
What does Codex Adversarial Review do?
A skill your agent uses to conduct aggressive, adversarial code reviews on git changes, branches, or PRs to detect critical edge cases, race conditions, security holes, and data loss risks. Codex Adversarial Review is an agent skill from WrongStack/WrongStack. Use this skill to conduct aggressive, adversarial code reviews on git changes, branches, or PRs to detect critical edge cases, race conditions, security holes, and data loss risks.
When should I use Codex Adversarial Review?
Codex Adversarial Review fits situations like: conduct aggressive; adversarial code reviews on git changes; PRs to detect critical edge cases; race conditions.
How do I install Codex Adversarial Review in Claude Code?
Run `npx skills add WrongStack/WrongStack --skill codex-adversarial-review -a claude-code`. Or copy the skill folder (packages/core/skills/codex-adversarial-review in WrongStack/WrongStack) into .claude/skills/codex-adversarial-review in your project. Claude Code loads it when a task matches its description.
How do I install Codex Adversarial Review in Codex?
Run `npx skills add WrongStack/WrongStack --skill codex-adversarial-review -a codex`. Or copy the skill folder (packages/core/skills/codex-adversarial-review in WrongStack/WrongStack) into .agents/skills/codex-adversarial-review in your project. Codex loads it when a task matches its description.
Can I use Codex Adversarial Review in Cursor, Gemini CLI or GitHub Copilot?
Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add WrongStack/WrongStack --skill codex-adversarial-review -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/codex-adversarial-review, .gemini/skills/codex-adversarial-review, .github/skills/codex-adversarial-review and .opencode/skills/codex-adversarial-review in your project.
What does Codex Adversarial Review need to run?
Going by SKILL.md and its folder, Codex Adversarial Review needs the command-line tools its instructions call (git).
Does Codex Adversarial Review access the network?
SKILL.md contains no URLs. Its commands use git, which can reach the network depending on how they are called. This is read from the text; nothing was executed.
Is Codex Adversarial Review safe to install?
Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.
What licence does Codex Adversarial Review use?
Codex Adversarial Review is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.
How many tokens does Codex Adversarial Review use?
About 2k tokens (SKILL.md is roughly 7.8k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.
What are the alternatives to Codex Adversarial Review?
Skills that share tags, products or a category with Codex Adversarial Review: Code Review Checklist (shareAI-lab/learn-claude-code, 78k stars), Requesting Code Review (HezaoHezao/poirot, 249 stars), Openqodex (openqodex/openqodex, 524 stars) and Code Review with Beads Tasks (maslennikov-ig/claude-code-orchestrator-kit, 260 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
Who maintains Codex Adversarial Review?
WrongStack (a GitHub organization) maintains it in WrongStack/WrongStack, which has 371 GitHub stars. The repository holds 100 skills in this directory. The repository was last updated on October 10, 2026.
Source: WrongStack/WrongStack on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.