dineway-content-ai-visibility
Pass
Audited by Gen Agent Trust Hub on Aug 15, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill is designed to process untrusted data from external AI service providers (indirect prompt injection surface). 1. Ingestion points: External AI responses and answer evidence collected via browser automation or agent clients (SKILL.md). 2. Boundary markers: No delimiters are specified to isolate untrusted external content. 3. Capability inventory: Writes reports to 'jobs/' and returns payloads to pipeline stages (SKILL.md). 4. Sanitization: No explicit sanitization or validation of the ingested AI responses is mentioned.
- [DATA_EXFILTRATION]: The skill performs network requests to external AI platforms. Evidence: Instructions specify interacting with services including ChatGPT, Gemini, and Claude (SKILL.md). Context: Network operations target well-known services and are core to the monitoring functionality.
Audit Metadata