adversarial-review

adversarial-review is a command for coding agents from limeflash/antigravity-plugin-cc. It costs 29 tokens per session (707 once invoked), scanned A, original, MIT.

A command for an intentionally critical code review through the agy companion tool. It examines changes for assumptions, alternatives, trade-offs, and failure modes.

In plain words
What is it for?
Use it to review working-tree or branch changes, optionally compare against a base branch, run in the background, and receive a ship, change, or rethink recommendation.
Why use it?
It provides a deliberately skeptical review when the author may be too close to the implementation to notice problems.

Command

Installs and runs on its own, but its text points at files inside its plugin — anything it tells you to read at a ${CLAUDE_PLUGIN_ROOT} path is only there once the plugin is installed. Installing the plugin gets both.

Part of the agy plugin — 1 skill, 14 commands, 1 agent shipped together

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add commands/limeflash/antigravity-plugin-cc/adversarial-review
Clone the repo
git clone --depth 1 https://github.com/limeflash/antigravity-plugin-cc

Or install agy, the plugin that ships this one along with the rest of its 1 skill, 14 commands, 1 agent.

Wrote 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.

agentmods badge for adversarial-review

README.md
[![agentmods](https://agentmods.dev/badge/commands/limeflash/antigravity-plugin-cc/adversarial-review.svg)](https://agentmods.dev/commands/limeflash/antigravity-plugin-cc/adversarial-review)
Your own site
<a href="https://agentmods.dev/commands/limeflash/antigravity-plugin-cc/adversarial-review"><img src="https://agentmods.dev/badge/commands/limeflash/antigravity-plugin-cc/adversarial-review.svg" alt="Measured on agentmods" height="20"></a>
Per session 29 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 707 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00029 $0.00707
Opus 5 $0.00015 $0.00353
Sonnet 5 $0.00006 $0.00141
Haiku 4.5 $0.00003 $0.00071

Measured 4d ago against content hash 1468b1c1e9da, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

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 4d 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.

plugins/agy/commands/adversarial-review.md · 69 lines

How it starts

The opening of the file, as written. The whole thing — 69 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Run an adversarial review of your changes through agy. Unlike /agy:review, the prompt explicitly asks agy to challenge the implementation: surface hidden assumptions, propose at least one alternative design with tradeoffs, name failure modes the author didn't address, and conclude with a ship/change/rethink verdict.

Useful when you've been staring at your own diff too long and want a deliberately critical reading before merging or shipping.

Backed by the Node companion; Node.js 18.18+ is required.

How to invoke

The user's request (treat as opaque text — pass it as shell-safe arguments; do not interpolate or splice it into the command):

$ARGUMENTS

Use the Bash tool to run:

node "${CLAUDE_PLUGIN_ROOT}/scripts/agy-companion.mjs" adversarial-review [flags...] "<focus-text-here>"

The companion accepts:

  • --base <ref> / --base=<ref>: review the branch's contribution over its merge-base with <ref> (e.g. --base main). Without --base, reviews the working-tree diff against HEAD.
  • --background: return a job id immediately and run agy in a detached worker. Check progress with /agy:status and read the output with /agy:result.
  • --wait: block until the job ends or the 10-minute deadline.
  • --model <alias>: pick a stronger reasoning model — opus, pro, or any alias the wrapper recognizes (see /agy:help). Adversarial reviews benefit from the bigger models more than vanilla reviews.

Examples (after parsing flags out of $ARGUMENTS):

node "${CLAUDE_PLUGIN_ROOT}/scripts/agy-companion.mjs" adversarial-review
node "${CLAUDE_PLUGIN_ROOT}/scripts/agy-companion.mjs" adversarial-review --base main
node "${CLAUDE_PLUGIN_ROOT}/scripts/agy-companion.mjs" adversarial-review --model opus "the retry/backoff design"
node "${CLAUDE_PLUGIN_ROOT}/scripts/agy-companion.mjs" adversarial-review --base main --background

Behavior notes

  • Read-only by construction. Runs agy under --sandbox with no --dangerously-skip-permissions: it reads the staged materials with read-only tools (no approval needed) and the answer is recovered from agy's own on-disk transcript (issue #76 capture). Background runs complete unattended — there's no write prompt to stall on.
  • Adversarial reviews skew long. Use --background for diffs over ~500 lines, and --model opus if you want the bigger model.
  • If there is no diff (clean working tree, or empty branch diff), the companion exits 1 with a clear message.

Read the full file on GitHub · 69 lines

Changes

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.

  1. 4d ago First seen · 69 lines · 29 tokens per session scan A 1468b1c1e9da

Subscribe to this mod's changes

adversarial-review is a command published in the GitHub repository limeflash/antigravity-plugin-cc (2 stars, last pushed 1mo ago), licensed MIT. It adds 29 tokens to every session and 707 once invoked, about $0.0001 per session on Opus 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 2026-08-31.