The Agent Skills Directory

[SAFE]: The skill provides documentation and examples for integrating with Zoom's official AI Services Translator API. All remote references and code examples point to trusted Zoom infrastructure (zoom.us, zoom.tech, github.com/zoom/*).
[CREDENTIALS_UNSAFE]: The skill correctly instructs developers to use environment variables (ZOOM_API_KEY, ZOOM_API_SECRET, AWS_ACCESS_KEY_ID, etc.) and explicitly warns to keep secrets server-side. Placeholders like AKIA... and secret are used in examples, which is acceptable practice.
[EXTERNAL_DOWNLOADS]: The skill mentions a quickstart sample from github.com/zoom/ai-services-quickstart/. This is a vendor-owned repository and is considered safe under the trust-scope-rule.
[COMMAND_EXECUTION]: One example uses curl to demonstrate a POST request to the Zoom API. This is a standard documentation pattern for API usage.
[OBFUSCATION]: No obfuscation, hidden content, or homoglyph attacks were detected in the instructions or code samples.
[INDIRECT_PROMPT_INJECTION]: While the skill processes user-supplied text for translation, it uses standard API request bodies. There is no evidence of unsafe interpolation that would lead to agent instruction override.

translator