Borrowing it
Nothing to install: this file belongs to nubbymong/claude-command-center. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/nubbymong/claude-command-center/main/.claude/skills/adversarial-review/SKILL.mdgit clone --depth 1 https://github.com/nubbymong/claude-command-centerWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/nubbymong/claude-command-center/adversarial-review)<a href="https://agentmods.dev/skills/nubbymong/claude-command-center/adversarial-review"><img src="https://agentmods.dev/badge/skills/nubbymong/claude-command-center/adversarial-review/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/nubbymong/claude-command-center/adversarial-review"><img src="https://agentmods.dev/badge/skills/nubbymong/claude-command-center/adversarial-review.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00136 | $0.04216 |
| Opus 5.5 | $0.00054 | $0.01686 |
| Sonnet 5.5 | $0.00027 | $0.00843 |
| Haiku 4.5 | $0.00014 | $0.00422 |
Grade A, and why
adversarial-review scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 297 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are the adversarial-review orchestrator. Your job is to make a change prove it cannot be
broken before it earns an approval -- by dispatching independent attackers, not by re-reading
the diff yourself. You are invoked directly (/adversarial-review <target>) or automatically when a
change touches a security-sensitive path.
INPUT (target): a PR number (#147/147), a commit range (origin/beta..HEAD), the working
diff, or a path set. If none given, default to the current branch's diff vs beta (this repo's
integration branch -- see AGENTS.md "Release Process").
EXECUTE. Do not ask, do not defer, do not offer.
This skill dispatches sub-agents immediately. When a change touches a security-sensitive path (Phase 0), the pass is required -- so running it is not a decision to put back to the caller.
Specifically, none of the following is a reason to stop and ask:
- "this fans out several agents" -- that is the mechanism, not a cost to be approved
- "the author can't review their own boundary" -- correct, which is why you dispatch others
- "the user might not want to spend the tokens" -- the cost is bounded by the budget rules below
- "I'll offer to run it" -- offering IS the failure mode this section exists to kill
The author orchestrating their own change is expected and fine. Writing the code then declining to dispatch the attackers leaves a required gate unrun and is the one outcome that must never happen. If the pass genuinely does not apply (docs-only, changelog-only, pure styling), say so in one line and stop -- that is the only early exit.
Model tiering (token efficiency is a design constraint)
Cheap models narrow the surface; expensive models attack the narrowed surface. Never hand a full diff to an expensive attacker.
| Phase | Model | Why |
|---|---|---|
| 0 -- scope + risk classification | fable |
mechanical: read --stat, match paths against the table |
| 0.5 -- thesis assertion check | fable |
mechanical: extract claimed guarantees, grep for the test that asserts each |
| 1 -- attackers | opus for the highest-severity lens, sonnet for the rest |
judgement work; this is where the budget goes |
| 2-5 -- synthesis, verdict, lifecycle | orchestrator | no extra agents |
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 1mo ago First seen · 297 lines · 136 tokens per session scan A 522f92d55416
adversarial-review is a skill published in the GitHub repository nubbymong/claude-command-center (57 stars, last pushed 17d ago), licensed MIT. It adds 136 tokens to every session and 4,216 once invoked, about $0.0005 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-15.
Other skills, from other repositories
merge-batch
Prepare and merge an owner-selected PR batch with contributor credit and exact-candidate admission. Use for a pasted PR list; combined focused validation is diagnostic, and every actual landing still requires its applicable checks and authorization.
make-pr-easy-to-review
Prepare PRs for review by cleaning noisy history, improving PR descriptions, and adding reviewer guidance without changing code behavior. Use for "make this easy to review", "tidy this PR", "clean up commits", or "annotate the diff".
thermo-nuclear-code-quality-review
Run an extremely strict maintainability review for abstraction quality, giant files, and spaghetti-condition growth. Use for a thermo-nuclear code quality review, thermonuclear review, deep code quality audit, or especially harsh maintainability review.
security-review
Perform a focused security review of pending git changes to identify high-confidence security vulnerabilities with real exploitation potential. Use this skill when the user asks for a security review, security audit, vulnerability scan, or wants to check pending changes on a branch for security issues before merging.…
huggingface-llm-trainer
Train or fine-tune language models with TRL or Unsloth on Hugging Face Jobs, including SFT, DPO, GRPO, reward models, and GGUF conversion. Use for cloud LLM training; use huggingface-vision-trainer for vision tasks.
review-work
Post-implementation gate review: run manual QA on the real surface yourself, then launch ONE gate reviewer (never a panel) to audit goal, constraints, code quality, security, missed context, and QA evidence. Use before a PR handoff or when the user explicitly asks to review completed work.