Borrowing it
Nothing to install: this file belongs to white1024/agent-nextup. 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/white1024/agent-nextup/main/.claude/skills/adversarial-review/SKILL.mdgit clone --depth 1 https://github.com/white1024/agent-nextupWrote 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/white1024/agent-nextup/adversarial-review)<a href="https://agentmods.dev/skills/white1024/agent-nextup/adversarial-review"><img src="https://agentmods.dev/badge/skills/white1024/agent-nextup/adversarial-review.svg" alt="Measured on agentmods" 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 | $0.00089 | $0.00717 |
| Opus 5 | $0.00044 | $0.00358 |
| Sonnet 5 | $0.00018 | $0.00143 |
| Haiku 4.5 | $0.00009 | $0.00072 |
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 6d 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 — 46 lines — stays where its author put it; the contents beside it link to each section on GitHub.
adversarial-review — adversarial document review
The principle: verification is not self-verification. The session that wrote the document cannot be its own reviewer. Dispatch a general-purpose fresh-context agent and tell it explicitly not to trust the author.
1. Set the scope
- With arguments: review those files.
- Without arguments: the
.mdfiles in the unpushed diff, including uncommitted changes. - Pure code changes are out of scope for this skill — that is what code review is for.
2. What the dispatch prompt must include
Instruct the agent to check each item against the code and the real environment, citing source file and line. Documents corroborating each other does not count.
| What to check | Check it against |
|---|---|
| UI wording (buttons, page names, fields, required vs optional) | Frontend source and the i18n resources |
| Commands and scripts | The project's build and dependency manifests, and whether the script files actually exist |
| Path claims (disk contract, output locations) | The path constants defined in code, and the real filesystem |
| List claims (tools, templates, commands) | The registry or enum in code (the single source of truth) |
| Behaviour claims (automation, defaults, the authorisation model) | The corresponding core module source |
| Relative links | Resolve every one of them |
| Internal consistency | Between the documents under review, and against existing outward-facing documents |
| The counting red line | Flag every tool count, test count or step count hard-coded in prose — those should point at a single source rather than being written out a second time |
If the project provides a "where to check each claim" reference (look at the development documents the CLAUDE.md index points to), include that whole table in the dispatch prompt so the agent does not have to find the sources itself.
Report format: one line per finding, [error/warn/nit] file:line · claim · reality (source file:line). List the categories that came back clean as "checked" too, and end with a verdict on whether this can be committed as is.
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.
- 6d ago First seen · 46 lines · 0 tokens per session scan A fed8b21f4b95
adversarial-review is a skill published in the GitHub repository white1024/agent-nextup (2 stars, last pushed 25d ago), licensed MIT. It adds 89 tokens to every session and 717 once invoked, about $0.0004 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.
Other skills, from other repositories
[object Object]
Skill "[object Object]" from microsoft/apm, covering what this package does, getting started, available primitives and included workflows.
apm-usage
Activate when the user asks about APM (Agent Package Manager): installing, configuring, authoring, or troubleshooting AI-agent packages, dependencies, compilation, MCP servers, policy, or any apm CLI command.
context-canvas-checkpoint
Maintain an optional session task map and explicit retrievable references in Context Canvas Codex. Use when navigation or long-context offload would materially help with goals, decisions, progress, dependencies, blockers, next steps, exploration summaries, large textual tool results, or selective historical…
context-canvas-reflection
Run one bounded trajectory reflection when repeated same-cause failures, contradicted assumptions, real-use failure after local green, material scope drift, a materially unresolved phase boundary, an unapproved authority-sensitive next effect, or user doubt suggests the current path may be wrong. Use Context Canvas…
operate-a2a-superhub
Operate and diagnose the A2A Superhub product across its CLI, HTTP, and declared MCP/A2A surfaces. Use when the user explicitly asks to inspect a Superhub, create or read Superhub tasks or artifacts, work with Superhub memory/inbox/wakeup/handoffs, validate this product skill, diagnose auth/index/queue/capability…
contorium-workspace-memory
Use Contorium workspace memory layers (focus, session events, Git, export) when resuming AI work or avoiding repeated project explanation.