agentsop-domain-eval-set
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill consists entirely of markdown-based procedural guidance and does not include any scripts, commands, or tools that interact with the host system or network.
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of real-world domain data, such as support tickets and logs (as specified in SKILL.md Stage 1), which could contain untrusted input. However, the risk is effectively mitigated by the skill's mandatory human labeling process (Stage 2) and inter-annotator agreement requirements, ensuring no adversarial content is processed without oversight. The skill lacks any code execution capabilities, further reducing the impact of potential injections.
- [SAFE]: The reference files include absolute file paths from the author's local development environment (e.g., in R1-source-evidence.md). These paths are environmental artifacts that do not reveal sensitive user data, credentials, or protected system configurations.
Audit Metadata