Skills
skills/tencent/ai-infra-guard/aig-agent-redteam/Gen Agent Trust Hub

aig-agent-redteam

Pass

Audited by Gen Agent Trust Hub on Jul 2, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill contains massive libraries of known jailbreak and prompt injection payloads (e.g., ChatGPT-Jailbreak-Prompts.json, JailBench-Tiny.json) which are used as test vectors to evaluate the safety boundaries of target AI models.
  • [EXTERNAL_DOWNLOADS]: The scripts/aig_data.py utility includes functionality to clone and synchronize security data from the official Tencent AI-Infra-Guard repository on GitHub to update vulnerability and fingerprint rules.
  • [COMMAND_EXECUTION]: The skill utilizes subprocess.run in scripts/aig_data.py to execute the git clone command, enabling the retrieval of essential security datasets required for infrastructure scanning.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 2, 2026, 07:11 AM
Security Audit — agent-trust-hub — aig-agent-redteam