The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes data from external webcam providers which could theoretically contain malicious instructions. However, the skill provides explicit, strong sanitization instructions under the 'Content safety' section, directing the agent to treat all responses strictly as data and ignore any directive-like text (e.g., 'ignore previous instructions').
[DATA_EXFILTRATION]: The skill uses a vendor-specific API endpoint (worldmonitor.app) for its primary functionality. It includes a placeholder for an API key (wm_0123456789...) as an example, which is a benign implementation detail. No sensitive local data access or unauthorized exfiltration patterns were detected.
[EXTERNAL_DOWNLOADS]: The skill references and fetches data from the WorldMonitor API and images from provider URLs (like Windy.com). These are legitimate, well-known services related to the skill's primary purpose.

monitor-webcams