Run a skeptical Codex review from Claude Code that hunts for hidden regressions, unsafe assumptions, and release risk. Use before merges, deploys, migrations, or other risky changes.
Scanned 9/3/2026
Install to Claude Code
Skills Directorynpx -y skills add IgorGanapolsky/ThumbGate --skill adversarial-review --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Adversarial Review?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/igorganapolsky-adversarial-review)More formats (shields.io, HTML) on the badges page.
---
name: adversarial-review
description: Run a skeptical Codex review from Claude Code that hunts for hidden regressions, unsafe assumptions, and release risk. Use before merges, deploys, migrations, or other risky changes.
---
# Adversarial Review
Default to the current uncommitted workspace unless the user explicitly provides `base=<branch>` or `commit=<sha>`.
Run one of:
```bash
node "${CLAUDE_PLUGIN_ROOT}/scripts/codex-bridge.js" adversarial-review --uncommitted
node "${CLAUDE_PLUGIN_ROOT}/scripts/codex-bridge.js" adversarial-review --base main
node "${CLAUDE_PLUGIN_ROOT}/scripts/codex-bridge.js" adversarial-review --commit <sha>
```
If the user asks for a specific threat model such as security, rollback safety, billing, or data loss, pass that in `--prompt`.
When reporting back:
- lead with concrete risks
- separate proven issues from speculation
- mention the saved artifact path if the user wants the full Codex output
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!
Adversarial multi-agent planning skill. Self-orchestrates 5 hostile category members (unspecified-low, unspecified-high, deep, ultrabrain, artistry) via team-mode for ruthless cross-critique debate, distills only the defensible insights, then MANDATORILY hands the distilled insight bundle to the `plan` agent for executable plan formalization. Use when planning needs maximum rigor and surfacing of weak assumptions, blind spots, and over-engineering. Triggers: 'hyperplan', 'hpp', '/hyperplan', ...
