negotiation
Pass
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: LOW
Full Analysis
- [SAFE] (SAFE): Comprehensive review of all 9 files reveals no malicious patterns. The skill is composed of purely textual coaching frameworks and reference material.
- [REMOTE_CODE_EXECUTION] (SAFE): No code execution commands (e.g., eval, exec, subprocess), script files (.py, .js, .sh), or package management files were detected.
- [DATA_EXFILTRATION] (SAFE): No network operation patterns (curl, wget, fetch) or sensitive file path access were identified.
- [PROMPT_INJECTION] (SAFE): Instructions follow standard agent role-definition patterns without attempting to bypass safety filters or override system constraints.
- [INDIRECT_PROMPT_INJECTION] (LOW): The skill is designed to process user-provided negotiation scenarios (untrusted data) to provide coaching. However, since the skill has no write, execute, or network capabilities, the risk of an injection causing side effects is negligible. It functions solely as a reasoning and text-generation tool.
Audit Metadata