Borrowing it
Nothing to install: this file belongs to robertoecf/adversarial-review. 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/robertoecf/adversarial-review/main/AGENTS.mdgit clone --depth 1 https://github.com/robertoecf/adversarial-reviewWrote 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/instructions/robertoecf/adversarial-review/agents-md)<a href="https://agentmods.dev/instructions/robertoecf/adversarial-review/agents-md"><img src="https://agentmods.dev/badge/instructions/robertoecf/adversarial-review/agents-md/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/instructions/robertoecf/adversarial-review/agents-md"><img src="https://agentmods.dev/badge/instructions/robertoecf/adversarial-review/agents-md.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 | $0.02234 | $0.02234 |
| Opus 5.5 | $0.00894 | $0.00894 |
| Sonnet 5.5 | $0.00447 | $0.00447 |
| Haiku 4.5 | $0.00223 | $0.00223 |
Grade B, and why
adversarial-review AGENTS.md scanned grade B with 1 finding 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 28d 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.
Reads agent configuration directoriesmediumAgent snooping
.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.
- **`forced_login_method = "chatgpt"`** must be in `~/.codex/config.toml` for How it starts
The opening of the file, as written. The whole thing — 157 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md: review agent guidelines (host-agnostic)
Plugin architecture
This plugin implements a single principle:
The partner reviews, never the host.
For every review (plan, code, or prompt — all through the single adversarial-review skill):
- The agent reading the SKILL.md is the host.
- The host must not review the host's own work.
- The host runs
lib/call-external.shto delegate the heavy critique to the OTHER agent (the partner). - The host then runs its own independent analysis and cross-validates
against the partner's output, tagging findings as
[cross-validated],[external-only], or[host-only].
There is no haiku courier subagent in this version. The host (main session) does both the external dispatch and the synthesis. Removed in the 0.5.0 refactor: the courier added complexity (model inconsistency across files, blocking-vs-non-blocking ambiguity) without proportionate value.
Cross-host routing
| Detected host | Partner |
|---|---|
claude |
Claude authors (default): T1 Astra high via Pi, then codex exec; T2 Pi chain; T3 Gemini. Codex authors: T1 is the Claude main loop inline; wrapper adds T2/T3, never Codex |
codex |
Codex/user/unknown authors: T1 claude -p; Claude/Pi/Grok authors: T1 Codex Astra. Then T2 Pi chain (non-xAI for Grok authors), T3 Gemini |
grok |
Codex Astra, then Gemini |
pi |
Codex Astra, then Claude, direct xAI, then Gemini |
unknown |
Direct xAI, then Pi, then Gemini |
Detection happens at every invocation via lib/detect-host.sh. See
references/host-detection.md for the priority order and the env-leak
asymmetry that drives "Codex env first, Claude env second".
Codex- and Claude-host callers should set ADVERSARIAL_REVIEW_AUTHOR to codex,
grok, claude, pi, user, or unknown. Aliases sol, openai, astra,
xai, and anthropic are accepted case-insensitively. On Codex, omitted authors
infer codex; user and unknown follow the same Grok-first route. On Claude,
omitted authors infer claude.
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.
- 28d ago Changed · +49 lines · +771 tokens per session scan A → B 28e867c72e4a
- 1mo ago First seen · 108 lines · 1,463 tokens per session scan A a6c399c6fe32
adversarial-review AGENTS.md is an instructions file published in the GitHub repository robertoecf/adversarial-review (12 stars, last pushed 28d ago), licensed MIT. It adds 2,234 tokens to every session, about $0.0089 per session on Opus 5.5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
vscode buildNext.instructions.md
Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).
next.js AGENTS.md
AGENTS.md instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.
vscode oss-third-party-notices.instructions.md
Instructions for microsoft/vscode, covering vs code oss third-party-notices pipeline, architecture, pipeline flow in ci, cache eligibility and rollout and applying the notice (cutover).
langchain AGENTS.md
AGENTS.md instructions for langchain-ai/langchain, covering global development guidelines for the langchain monorepo, corridor security analysis, project architecture and context, monorepo structure and development tools & commands.
vscode copilot-instructions.md
Copilot instructions for microsoft/vscode, covering vs code copilot instructions, project overview, root folders, core architecture (src/ folder) and built-in extensions (extensions/ folder).
deepseek-harness AGENTS.md
AGENTS.md instructions for deepseek-ai/deepseek-harness, covering agents.md, pre-stable apis and released session data, repository layout, commands and host sandbox failures.