RepoAI
Browse MCPs

AIG Agent Redteam

100TrustedShips scripts

Conduct unauthorized-access simulations for AI agents and infrastructure.

What it does

This skill equips an assistant to perform authorized security assessments on AI products, agents, and their underlying code or infrastructure. By focusing on first-principles reasoning rather than static payload execution, the agent maps potential trust boundaries, constructs attack scenarios, and adapts its testing strategy based on real-time feedback. It ensures that every security finding is supported by concrete evidence rather than automated guesses.

Designed for detailed risk identification, the skill automates the evidentiary heavy lifting—such as fingerprinting, data aggregation, and report generation—while leaving the critical safety judgments to the AI. It provides a structured way to report vulnerabilities in business terms, outlining specific consequences and actionable remediation steps without relying on destructive actions.

When it earns its place

Reach for it when…

Apply this when you need to audit an AI agent, MCP server, or software supply chain for risks like prompt injection, unauthorized tool use, data leakage, or SSRF. It is ideal for pre-deployment validation or regular security posture checks where you need a documented, evidence-based report of how an agent handles adversarial inputs or attempts to cross defined security boundaries.

Look elsewhere when…

Do not use this for automated, high-volume vulnerability scanning against public services without explicit written authorization, as it is a testing framework rather than a penetration testing tool for black-box exploitation. It is also not suitable for replacing professional security audits, as it requires human oversight to define the scope and verify that the proposed testing boundaries are strictly maintained.

What to say to use it

You never invoke a skill directly — the assistant picks it up when your request matches what it covers. These are the kinds of things that do that.

  • Test this agent for prompt injection vulnerabilities.
  • Perform a security assessment on this MCP server.
  • Can you run a red team drill on my current configuration?
  • Generate a penetration testing report for this agent.

Install it

Install for all your projects

Copy the skill directory into your personal skills folder. Claude picks it up on its own the next time your request matches what it does — there is nothing to enable.

bash
git clone --depth 1 --branch main https://github.com/Tencent/AI-Infra-Guard.git /tmp/AI-Infra-Guard
mkdir -p ~/.claude/skills
cp -r /tmp/AI-Infra-Guard/skills/aig-agent-redteam ~/.claude/skills/aig-agent-redteam

Install for one project only

Put it in the project's own skills folder instead, so it can be committed and shared with everyone working on that repository.

bash
cp -r /tmp/AI-Infra-Guard/skills/aig-agent-redteam .claude/skills/aig-agent-redteam

If a skill of the same name exists in both places, the personal one wins.

Installing on Claude.ai or through the API instead? The full guide covers every surface.

Pair it with an MCP server

AIG Agent Redteam changes how the work is done. These servers change what Claude can reach while doing it.

How skills and MCP servers differ →