multimodal-llm
Pass
Audited by Gen Agent Trust Hub on Apr 18, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides instructional patterns for integrating with well-known AI services such as OpenAI, Google Gemini, Anthropic, AssemblyAI, Kling AI, and fal.ai. All API interactions are documented using established protocols (REST, WebSockets) and official or widely-used SDKs.
- [SAFE]: Network operations described in the skill rules target legitimate technology providers and cloud services. The use of WebFetch and WebSearch tools is appropriately scoped to the stated purpose of processing and retrieving multimodal content.
- [SAFE]: Hardcoded credential security is maintained throughout the documentation by using standard placeholders (e.g., 'YOUR_API_KEY', 'your-api-key') and advising the use of environment variables for secret management.
- [SAFE]: The skill documentation correctly identifies and mitigates common integration risks, such as implementing exponential backoff for polling asynchronous video generation tasks and handling large PDF context windows via page-range processing.
Audit Metadata