skilld
All skills
tencent avatar

/aig-agent-redteam

@06b9eb9
by tencenttencent/ai-infra-guard6.6k stars
622

当用户要求 AI/Agent 安全评估、蓝军演习、AI 安全审查、提示词注入测试、MCP/Skill/插件/代码包审计、Agent 工具链滥用测试，或需要生成类似渗透测试报告的 Markdown/HTML 时，必须使用本 skill。本 skill 让 Agent 以授权蓝军视角成为 AI 安全专家，面向 AI 产品、Agent、MCP Server、Skill、代码仓库和 AI 基础设施进行安全演习。优先使用第一性原理推理和真实证据，而不是机械跑 payload 库；脚本只用于 HTTP 指纹识别、证据聚合、报告渲染等确定性辅助任务。

Ask your Agent

Use this Skill: https://skilld.dev/gh/tencent/ai-infra-guard/aig-agent-redteam

This session only. Nothing lands on disk.

README.md

≈635 tokens on demand. Your agent reads this file only when SKILL.md points to it.

A.I.G Agent Red Team — Agent Security Assessment in One Command

中文版

A comprehensive AI Agent red-team security assessment skill. Install into your Agent client and let it attack itself — testing prompt injection, indirect injection, tool abuse, data leakage, privilege escalation, SSRF, supply chain risks, and infrastructure exposure.

Powered by first-principles red-team methodology from Tencent Zhuque Lab A.I.G.

Install

npx skills add https://github.com/Tencent/AI-Infra-Guard.git --skill aig-agent-redteam

Alternatively, install from source:

https://github.com/Tencent/AI-Infra-Guard/tree/main/skills/aig-agent-redteam

Requires node and git. Supports Claude Code, CodeBuddy, Cursor, and other Agent clients.

If npx skills is unavailable, install manually:

git clone https://github.com/Tencent/AI-Infra-Guard.git /tmp/aig
cp -r /tmp/aig/skills/aig-agent-redteam ~/.claude/skills/

Usage

After installation, just chat:

Personal Developer (IDE Agent) — audit installed Skills, MCP Servers, and supply chain:

帮我进行安全演习

Production Agent — full attack surface coverage: infrastructure scan, code audit, dynamic testing, jailbreak evaluation, workflow attack:

帮我进行安全演习

The skill automatically identifies the agent type and adapts the test scope. For production agents, it covers:

  • Infrastructure Scan: fingerprint + CVE matching for AI services (Ollama, vLLM, Dify, etc.)
  • Code Audit: static analysis of Skill source, MCP Server code, and dependency supply chain
  • Dynamic Testing: 30+ prompt injection payloads with adaptive mutation (role-play, encoding, multi-turn escalation)
  • Jailbreak Evaluation: LLM boundary testing via Parseltongue encoding engine
  • Workflow Attack: multi-step task chain abuse, indirect injection via documents/RAG
  • Full Report: severity-rated findings with evidence chains, defense validation, and remediation advice — all data stays local

Notes

  • All test data stays on your machine; no data is uploaded to any server.
  • The skill uses first-principles reasoning, not blind payload execution.
  • Target scope must be confirmed before execution; destructive actions require explicit authorization.
  • Compatible with Claude Code, CodeBuddy, Cursor, and any Agent client supporting skills.

Powered by Tencent Zhuque Lab A.I.G

Source: SKILL.md on GitHub

Third-party checks

2 alerts1mo3 checks · Risk SAFE
  • Gen Agent Trust Hub1mo

    This is a specialized security assessment (red-teaming) tool from Tencent Zhuque Lab designed to audit AI agents and infrastructure. It includes prompt injection and obfuscation techniques as core features for security testing.

  • Socket1mo

    43 alerts: gptSecurity, gptAnomaly, gptMalware

  • Snyk1mo

    Risk: CRITICAL · 3 issues

Provenance

Signed by skilld at 06b9eb9. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub yesterday.

Activeupdated last month

Capability

version
5.0.0
Other metadata
metadata
{
  "author": "Tencent Zhuque Lab",
  "repo": "https://github.com/tencent/AI-Infra-Guard",
  "license": "Apache-2.0"
}

README badge

README badge for tencent/ai-infra-guard/aig-agent-redteam