containment
Pass
Audited by Gen Agent Trust Hub on Jul 20, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill includes a dedicated 'Prompt Injection Safety Notice' in Section 8. This section provides explicit instructions to the AI agent to disregard any commands or directives found within the analyzed data, such as command-and-control strings or phishing content. This proactive measure effectively mitigates risks associated with indirect prompt injection.
- [DATA_EXFILTRATION]: No network tools or commands (e.g., curl, wget, fetch) are present in the skill. The execution environment is restricted to file system operations through the 'allowed-tools' metadata (Read, Grep, Glob), and the skill explicitly forbids the inclusion of full attacker indicators in the final output.
- [COMMAND_EXECUTION]: The skill does not contain any shell execution patterns, persistence mechanisms, or privilege escalation attempts. It specifically instructs the agent that it must never execute containment actions directly and that all outputs are for human execution only.
- [SAFE]: The skill has no external dependencies, does not download remote scripts, and contains no obfuscated content. The references provided are to legitimate cybersecurity frameworks and industry-standard documentation.
Audit Metadata