monitor-webcams
Pass
Audited by Gen Agent Trust Hub on Sep 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill demonstrates safe behavior and follows best practices for API integrations. It correctly guides the user to manage API keys using environment variables rather than hardcoding them.
- [INDIRECT_PROMPT_INJECTION]: The skill proactively addresses potential security risks associated with processing external data. The 'Content safety' section provides clear guidance for the agent to distinguish between data and instructions, preventing the execution of malicious directives that might be embedded in webcam metadata.
- Ingestion points: External data is retrieved from the
worldmonitor.appAPI via thelist-webcamsandget-webcam-imageendpoints (SKILL.md). - Boundary markers: Explicit instructions are provided in the 'Content safety' section to disregard directive-like text within responses.
- Capability inventory: The skill uses standard network tools (
curl) to interact with the API (SKILL.md). - Sanitization: The skill employs prompt-based safeguards to ensure the agent ignores potential injection attempts in the ingested data.
Audit Metadata