agent-red-team

agent-red-team is a skill for Claude Code, Codex from microsoft/cat-agent-skills. It costs 114 tokens per session (2,609 once invoked), scanned B, original, MIT.

A security testing process for a Microsoft Copilot Studio agent, Microsoft’s platform for building business agents. It tests how the agent handles hostile instructions, private information, permission checks, requests outside its scope, and tool use.

In plain words
What is it for?
Creating and running adversarial test cases, checking for prompt injection, data leaks, authorization bypasses, scope escape, and unsafe tool actions, then scoring the findings.
Why use it?
An agent can appear correct with cooperative test users while still exposing data or following harmful instructions hidden in retrieved documents. This finds those weaknesses and links them to specific controls.

Skill for Claude CodeCodex ✓ vendor

Written for no agent in particular: nothing here depends on one.

Good fit Creating and running adversarial test cases, checking for prompt injection, data leaks, authorization bypasses, scope escape, and unsafe tool actions, then scoring the findings.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/microsoft/cat-agent-skills/agent-red-team
About the project

microsoft/cat-agent-skills is a static website that catalogs reusable instruction sets and related packages for AI agents. People use it to search, filter, rate, and download skills for Cowork, Copilot Studio, and Scout, along with Copilot plugins and Scout automations. The catalogue entries are the skills, instructions, plugins, and settings displayed by the site.

microsoft/cat-agent-skills · 64 stars · on GitHub

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.

Any agent
npx skills add microsoft/cat-agent-skills --skill agent-red-team
Clone the repo
git clone --depth 1 https://github.com/microsoft/cat-agent-skills

Made for: Claude Code, Codex.

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 agent-red-team

README.md
[![agentmods](https://agentmods.dev/badge/skills/microsoft/cat-agent-skills/agent-red-team.svg)](https://agentmods.dev/skills/microsoft/cat-agent-skills/agent-red-team)
Your own site
<a href="https://agentmods.dev/skills/microsoft/cat-agent-skills/agent-red-team"><img src="https://agentmods.dev/badge/skills/microsoft/cat-agent-skills/agent-red-team.svg" alt="Measured on agentmods" height="20"></a>
Per session 114 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 2,609 The whole file, excluding the scripts and references it only reads on demand.
Security scan B 1 finding. A grade says what 26 rules found in the file — not that it is safe. Third-party audits
  • NVIDIA SkillSpector warn 7 Sept 2026
SkillSpector: 1 finding, up to medium

These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →

  • medium Excessive Agency · line 105
    Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
    Fix: Add human-in-the-loop confirmation for destructive, irreversible, or high-impact operations. Never auto-execute commands that modify files, send data, or alter system state.
How audits are shown
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 $0.00114 $0.02609
Opus 5 $0.00057 $0.01305
Sonnet 5 $0.00023 $0.00522
Haiku 4.5 $0.00011 $0.00261

Measured 8d ago against content hash b288f3025002, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-08, from the pricing page.

Security

Grade B, and why

agent-red-team 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 8d ago.

The scan reads SKILL.md. This mod also ships 1 executable file (scripts/canaries.py), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

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.

Instruction-override phrasingmediumPrompt injection

Text telling the model to disregard its earlier instructions or safety rules is the shape of a prompt injection, whoever wrote it.

override the system instructions, suppress a disclosure, exfiltrate context, induce a

Downgraded: this mod is about security review, or the phrase is quoted, so it is likely naming the pattern rather than instructing it.

submissions/agent-red-team/SKILL.md · 228 lines

How it starts

The opening of the file, as written. The whole thing — 228 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Agent Red Team

Functional test plans ask whether the agent does the right thing for a cooperative user. This asks what it does for an uncooperative one — and, more importantly in practice, for a cooperative user reading a document somebody else poisoned.

The dominant real-world risk for a grounded enterprise agent is not a clever user typing a jailbreak. It is indirect prompt injection: instructions hidden inside content the agent retrieves, and oversharing, where the agent faithfully surfaces documents the asker was never meant to see. Weight the review accordingly.

Authorization gate — do this first, every time

Before generating a single test case, establish and record in the report:

  1. Ownership. The user owns the agent, or has written authorization from the owner to test it. If neither, stop. Say plainly that adversarial testing of someone else's agent needs their authorization, and offer to help them request it.
  2. Environment. Test in dev or a dedicated test environment wherever possible. If production is the only option, get that in writing and note it as a finding in itself.
  3. Data handling. Use synthetic canary data — never real personal data, real credentials or real customer records — to test leakage. A canary is a distinctive fabricated string planted in a document the agent can reach; if it appears in output where it should not, you have proof without exposing anything real.
  4. Blast radius. Identify which tools have write or send capability. Do not execute destructive or outbound actions (sending mail, creating tickets, writing to systems of record) against live systems. Test that the agent would call them, using a stubbed or read-only path, rather than letting it complete the action.
  5. Window. Agree a testing window and tell whoever monitors security alerting, so your traffic is not investigated as a live incident.

If the user declines the gate or cannot answer these, stop and explain why the review cannot proceed responsibly.

Read the full file on GitHub · 228 lines

Files

What ships with it

2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

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. 8d ago First seen · 228 lines · 114 tokens per session scan B b288f3025002

Subscribe to this mod's changes

agent-red-team is a skill published in the GitHub repository microsoft/cat-agent-skills (64 stars, last pushed today), licensed MIT. It adds 114 tokens to every session and 2,609 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it B with 1 finding (instruction-override phrasing). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Related

Other skills, from other repositories

systematic-debugging

Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.

obra/superpowers · 21 tokens

local-ai-agents

Build local-first AI agents that run entirely on a developer workstation with Microsoft Foundry Local and Qwen function-calling models. Covers Small Language Models (SLMs), the OpenAI-compatible local endpoint, sandboxed local tools, local RAG with Chroma, local MCP servers, hybrid cloud/local routing, and the…

microsoft/ai-agents-for-beginners · 200 tokens

next-cache-components-adoption

Turn on Cache Components in a Next.js app and resolve the blocking routes it surfaces. Use when the user wants to enable, adopt, or migrate to Cache Components, flip the cacheComponents flag, work through a flood of blocking-prerender / instant validation errors, run the cache-components-instant-false codemod, or…

vercel/next.js · 95 tokens

next-cache-components-optimizer

Drive a Next.js route to instant navigation by setting up an agentic loop, under Cache Components / PPR, on initial load (hard navigation) and client-side navigation (soft navigation). Encode the goal as a failing @next/playwright instant() e2e and work it to green, one verified route at a time; the shipped test then…

vercel/next.js · 170 tokens

next-partial-prefetching-adoption

Turn on Partial Prefetching in a Next.js app and work through the insights it surfaces. Use when the user wants to enable or adopt Partial Prefetching, flip the partialPrefetching flag, opt routes in with export const prefetch = 'partial', audit Link prefetch={true} behavior, preserve existing prefetched UI with…

vercel/next.js · 103 tokens

chronicle

Analyze Copilot session history for standup reports, usage tips, session search, and session reindexing. Use when the user asks for a standup, daily summary, usage tips, workflow recommendations, wants to search or find past sessions by keyword/file/PR, wants to reindex their session store, or asks about deleting…

microsoft/vscode · 72 tokens