role-red-team-critique

role-red-team-critique is a skill for Claude Code from chf3198/copilot-governance. It costs 70 tokens per session (1,563 once invoked), scanned A, original, MIT.

An adversarial review process that sends project records and governance documents to an independent model from a different model family. It returns structured findings and an ACCEPT, REJECT, or PARTIAL verdict.

In plain words
What is it for?
Use it to review handoff records, pull requests, governance documents, instructions, scripts, and deployed code changes, especially before key delivery handoffs.
Why use it?
It adds a second opinion that can expose omissions, weak evidence, or mistaken assumptions missed by the primary review.

Skill for Claude Code

Written for Claude Code: disable-model-invocation in frontmatter. Also seen: reads .claude/ paths; mentions CLAUDE.md; mentions Claude Code.

Needs its repository: it runs a file that does not travel with it, so clone the repository first. The line is node scripts/global/fleet-red-team-dispatch.js \.

Good fit Use it to review handoff records, pull requests, governance documents, instructions, scripts, and deployed code changes, especially before key delivery handoffs.

Compare 6 skills from other repositories ↓
Install

Getting it into your agent

It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.

Clone the repo
git clone --depth 1 https://github.com/chf3198/copilot-governance
agentmods
npx agentmods add skills/chf3198/copilot-governance/role-red-team-critique

Made for: Claude Code.

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 role-red-team-critique

README.md
[![agentmods](https://agentmods.dev/badge/skills/chf3198/copilot-governance/role-red-team-critique/github.svg)](https://agentmods.dev/skills/chf3198/copilot-governance/role-red-team-critique)
Your own site
<a href="https://agentmods.dev/skills/chf3198/copilot-governance/role-red-team-critique"><img src="https://agentmods.dev/badge/skills/chf3198/copilot-governance/role-red-team-critique/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.

agentmods 80×15 button for role-red-team-critique

Your own site · 80×15
<a href="https://agentmods.dev/skills/chf3198/copilot-governance/role-red-team-critique"><img src="https://agentmods.dev/badge/skills/chf3198/copilot-governance/role-red-team-critique.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 70 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,563 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
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 5.1 $0.00070 $0.01563
Opus 5 $0.00035 $0.00781
Sonnet 5 $0.00014 $0.00313
Haiku 4.5 $0.00007 $0.00156

Measured 11d ago against content hash 905fd6ee21ac, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-11, from the pricing page.

Security

Grade A, and why

role-red-team-critique 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 11d 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.

skills/role-red-team-critique/SKILL.md · 153 lines

How it starts

The opening of the file, as written. The whole thing — 153 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Role: Red-Team Critique

Purpose

Red-Team critique provides an adversarial second opinion on baton artifacts and governance documents by routing to a cross-family fleet model. The cross-family invariant (non-Anthropic reviewer) eliminates same-family echo-chamber effects where the primary model overlooks its own blind spots.

This skill is the canonical umbrella over:

  • scripts/global/fleet-red-team-dispatch.js — HAMR-wrapped Ollama dispatcher
  • scripts/global/baton-fleet-review-comment.js — structured comment formatter
  • scripts/global/red-team-evidence-quality.js — validator for evidence blocks
  • .claude/commands/fleet-review.md — slash command entry point
  • config/red-team-model-matrix.yml — A1 model-selection function

Dispatch Protocol

When to invoke

Red-Team is ADVISORY on all lanes. Invoke before posting the-collaborator-handoff on tickets labeled area:governance, area:instructions, or area:scripts. For lane:code-change touching deployed runtime artifacts, invoke before the-admin-handoff. Invocation is always optional; skipping must be noted in the-collaborator-handoff as red_team_review: skipped — <rationale>.

How to invoke

# High-stakes review (32B model, ~3-5 min)
node scripts/global/fleet-red-team-dispatch.js \
  --artifact-type collaborator-handoff \
  --ticket 2317 \
  --model qwen2.5-coder:32b

# Fast fallback (7B model, ~30-60 sec)
node scripts/global/fleet-red-team-dispatch.js \
  --artifact-type collaborator-handoff \
  --ticket 2317 \
  --model qwen2.5-coder:7b

Model selection at dispatch time uses config/red-team-model-matrix.yml via selectModel(taskContext, opts) in fleet-red-team-dispatch.js. Pass --stakes high to force 32B; omit for automatic selection per the matrix.

Cross-family invariant

The dispatched model MUST differ from the primary authoring model family. Claude Code sessions use Anthropic models; therefore all red-team dispatches MUST target Ollama-hosted models (Qwen, Llama, Mistral, etc.) or an OpenRouter non-Anthropic provider. Dispatching to another Claude model violates the cross-family rule and produces findings that must be discarded.

Read the full file on GitHub · 153 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. 11d ago First seen · 153 lines · 70 tokens per session scan A 905fd6ee21ac

Subscribe to this mod's changes

role-red-team-critique is a skill published in the GitHub repository chf3198/copilot-governance (2 stars, last pushed 1mo ago), licensed MIT. It adds 70 tokens to every session and 1,563 once invoked, about $0.0003 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.