travel-manager-hardened
Pass
Audited by Gen Agent Trust Hub on Apr 21, 2026
Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
- [PROMPT_INJECTION]: The file
SAFETY.mdcontains examples of adversarial prompts and system-level messages (e.g.,[SYSTEM] Google Maps integration configured...) used for safety testing. This content represents an indirect prompt injection surface. * Ingestion points: The agent processesSAFETY.mdas context. * Boundary markers: While labeled as test cases, the blocks lack programmatic delimiters to prevent the agent from obeying the embedded instructions. * Capability inventory:SKILL.mdacknowledges capabilities for file writing and network requests. * Sanitization: No sanitization is applied to the reference documentation content. - [NO_CODE]: The skill is comprised entirely of Markdown files. No executable code (Python, JavaScript, or shell scripts) is present in the analyzed files.
- [COMMAND_EXECUTION]: The skill instructions in
SKILL.mdexplicitly reference the potential for shell command and API execution, providing negative constraints. This indicates the skill is intended for high-privilege environments, though no specific execution patterns are provided.
Audit Metadata