research-engineer
Pass
Audited by Gen Agent Trust Hub on Apr 11, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill uses assertive language to define a specific persona ("You are not an assistant", "No Emojis. No Pleasantries"), but these instructions are intended to shape the agent's professional tone and interaction style rather than bypass safety filters or ignore core system instructions.
- [DATA_EXFILTRATION]: No sensitive file access, credential usage, or unauthorized network operations were identified.
- [REMOTE_CODE_EXECUTION]: The skill provides static code examples (C++, Python) for illustrative purposes and does not perform any remote code downloads or automated package installations.
- [COMMAND_EXECUTION]: No shell commands, privilege escalation attempts, or persistence mechanisms are present in the skill instructions.
- [SAFE]: The skill primarily consists of natural language instructions for a research-oriented persona and does not contain any executable scripts or dangerous configurations.
Audit Metadata