check-airport-delays
Pass
Audited by Gen Agent Trust Hub on Nov 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from the WorldMonitor API, which is an external ingestion point. The skill correctly identifies this as a potential attack surface and includes a 'Content safety' section with explicit instructions for the agent to treat all retrieved content as data rather than instructions.\n
- Ingestion points: API responses from the
worldmonitor.appendpoint (SKILL.md).\n - Boundary markers: A dedicated 'Content safety' section tells the agent to disregard directive-like text such as 'ignore previous instructions' or 'run this command' if found in response fields.\n
- Capability inventory: The skill performs network requests and data processing using
curlandjq.\n - Sanitization: Prompt-based sanitization is used to ensure the agent maintains its role and does not follow instructions embedded in data fields.\n- [PROMPT_INJECTION]: A heuristic detection for 'ignore previous instructions' was triggered; however, this is a false positive. The text is used defensively to instruct the agent on how to handle malicious payloads in external data, not to subvert the agent's own programming.
Audit Metadata