decisions
Pass
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted text and image input to perform classification and routing tasks, creating a surface for indirect prompt injection.\n
- Ingestion points: Processes user-provided text and image content via the
inputandinput_imagefields inSKILL.mdandreferences/api-contract.md.\n - Boundary markers: The skill relies on structured JSON but does not define specific text delimiters to isolate untrusted user content from the classification intent.\n
- Capability inventory: Executes network requests to
api.zoom.usto drive application-level decisions such as ticket escalation and queue routing.\n - Sanitization: Includes robust security guidance, specifically warning against fetching arbitrary URLs to prevent SSRF and advising that AI outputs must not serve as the sole basis for security or authorization boundaries.
Audit Metadata