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.
npx skills add metago-ai/metagolifeform --skill metago-adversarial-reviewgit clone --depth 1 https://github.com/metago-ai/metagolifeformWrote 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/metago-ai/metagolifeform/metago-adversarial-review)<a href="https://agentmods.dev/skills/metago-ai/metagolifeform/metago-adversarial-review"><img src="https://agentmods.dev/badge/skills/metago-ai/metagolifeform/metago-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/metago-ai/metagolifeform/metago-adversarial-review"><img src="https://agentmods.dev/badge/skills/metago-ai/metagolifeform/metago-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 | $0.00099 | $0.01480 |
| Opus 5 | $0.00049 | $0.00740 |
| Sonnet 5 | $0.00020 | $0.00296 |
| Haiku 4.5 | $0.00010 | $0.00148 |
Grade A, and why
metago-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 8d 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.
What it actually says
对抗式审查（metago-adversarial-review）
描述
凡任务属于测试 / 检查 / 审计 / 验证四类（含代码审查、安全审计、数据校验、交付验证、回归测试等），必须切换为对抗者视角执行：假设自己是攻击者/最苛刻的审查者，主动寻找会导致系统失效、数据泄露、逻辑错误的致命缺陷。禁止自证式检查、禁止把"看起来没问题"当作"没问题"。基于 A38 对抗式审查公理 + D45 对抗式审查强制触发属性。
触发条件
测试 / 检查 / 审计 / 验证类任务自动触发（A38 · D45 强制）。4.8 运行协议；与 metago-delivery-gate、metago-code-review-deep、metago-security-audit 协同。
⚠️ 强制说明（不依赖触发词）：本技能由 A38 公理 + D45 属性无条件强制激活——用户未说出任何触发关键词，AI 也必须对所有测试/检查/审计/验证类任务执行对抗式审查（角色切换为攻击者 → 威胁建模 → 致命缺陷挖掘 → 红队报告 → 修复闭环），跳过即违反 A38。下方 YAML
trigger:列表仅是"用户主动点名"时的快捷通道，不是激活前提。判定：验证/审计报告无"对抗式审查结论"小节 = 违规。
核心流程（六步）
-
角色切换（Switch Role）
- 立即从"实现者/检查者"切换为"攻击者/最苛刻的审查者"
- 心态：我的职责是让这个系统失败，而不是确认它正常
-
威胁建模（Threat Modeling）
- 假设有人要利用本系统/产物作恶或让它失败
- 识别：攻击者是谁（外部/内部/恶意用户）？目标是什么？入口在哪？
-
攻击面枚举（Attack Surface）
- 列出所有可被利用的入口：输入、接口、权限边界、依赖、配置、默认值、错误路径
- 对每个入口追问：最坏情况下会发生什么？
-
致命缺陷挖掘（Find Critical Flaws）
- 主动寻找会导致：系统失效 / 数据泄露 / 逻辑错误 / 越权访问 / 资源耗尽的缺陷
- 优先攻击：边界条件、异常输入、未授权路径、竞态、超时、降级逻辑
- 每个缺陷必须有证据（复现步骤 / 输入 / 输出）
-
红队报告（Red Team Report）
- 输出必须包含"对抗式审查结论"小节：
- 角色视角声明
- 攻击面清单
- 致命缺陷（严重度分级 + 证据）
- 已确认安全项（简要）
- 输出必须包含"对抗式审查结论"小节：
-
修复闭环（Fix Loop）
- 发现的缺陷必须：立即修复，或明确降级为"已知风险"（记录原因 + 缓解措施）
- 禁止忽略、禁止"以后再说"、禁止"用户不会那样用"
输出格式
## 对抗式审查结论（A38）
- 视角：攻击者 / 最苛刻审查者
- 攻击面：[入口1, 入口2, ...]
- 致命缺陷：
- [严重度] 缺陷描述 — 证据（复现步骤）
- 修复闭环：[已修复 / 已降级为已知风险（原因+缓解）]
- 已确认安全项：[...]
判定标准（通过/不通过）
- ✅ 通过：完成攻击面枚举 + 致命缺陷挖掘（含证据）+ 修复闭环，报告中含"对抗式审查结论"小节
- ❌ 不通过：自证式检查（"功能正常"）、无攻击面、无缺陷证据、缺陷未闭环 → 禁止宣告完成
与其他技能的协同
- 与
metago-delivery-gate协同：L9 对抗式审查为交付门禁必填小节 - 与
metago-code-review-deep协同：代码审查结果必须以对抗视角复核 - 与
metago-security-audit协同：安全审计即对抗式审查在安全域的具体化 - 与
metago-critique协同：批判分析侧重逻辑/事实，对抗式审查侧重攻防/失效 - 与
metago-first-principles协同：F0 事实基元作为对抗式审查的验证基准
根源文档
《元构全息智能引擎》（A38 对抗式审查公理、D45 对抗式审查强制触发属性、4.8 运行协议、13.2 第 12 维度、14.2 L9 小节）
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.
- 8d ago First seen · 85 lines · 99 tokens per session scan A 69f8bd691c97
metago-adversarial-review is a skill published in the GitHub repository metago-ai/metagolifeform (4 stars, last pushed 10d ago), licensed MIT. It adds 99 tokens to every session and 1,480 once invoked, about $0.0005 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 2026-09-03.
Other skills, from other repositories
vue-best-practices
A review checklist for Vue 2 and Vue 3 components, covering component names, prop declarations, TypeScript casts, API styles and complex template expressions.
plugin-quality
A review guide for checking Zhin.js plugins before release. It covers plugin structure, feature declarations, resource cleanup, message sending, and security.
rust-review
A Rust code-review skill for services, checking for crashes, unsafe SQL construction, exposed credentials, ignored errors, and unfinished code.
github-pr-comment
Post one exact GitHub pull-request comment through local gh, any compatible hosted connector, or the explicit human-gated MCP composition, with retry safety and independent readback.
review-skill
Inspect, safely test, and assess one Runx skill package for capability, trust, and operator readiness. Use when deciding whether to adopt, improve, reject, install, or publish a skill; its evidence-only assess runner is available when native test evidence already exists.
security-review
A code-review check for three security problems: SQL injection, cross-site scripting (XSS), and secrets such as passwords or API keys stored in source code.