reversing-network
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted network traffic (headers, body) from external websites via the
browser_get_request_responsetool. This data could contain malicious instructions designed to influence the agent's analysis or logic during the flow generation process. \n - Ingestion points:
browser_get_request_response(SKILL.md) \n - Boundary markers: None specified for the analysis of captured traffic. \n
- Capability inventory:
Core.Net.HttpRequest(network access),Core.Programming.Function(dynamic execution). \n - Sanitization: No specific sanitization or validation of the captured network data is mentioned before it is processed by the agent.\n- [DYNAMIC_EXECUTION]: The provided code templates utilize
Core.Programming.Functionnodes, which execute JavaScript logic provided as strings in thefuncproperty. While standard for the Robomotion SDK, this represents a dynamic execution pattern for data transformation tasks.
Audit Metadata