ai-model-gateway
Pass
Audited by Gen Agent Trust Hub on Jun 15, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides technical architecture and implementation guidance without any malicious scripts, persistence mechanisms, or unauthorized data access patterns.
- [EXTERNAL_DOWNLOADS]: The documentation references standard, official SDKs and APIs from well-known AI service providers (Anthropic, OpenAI, Google, and AWS Bedrock) for legitimate integration purposes.
- [PROMPT_INJECTION]: While the gateway processes untrusted user variables, the architecture explicitly includes a 'Safety In' pipeline stage with prompt-injection classifiers and a 'Safety Out' stage for output scrubbing and grounding checks to mitigate indirect prompt injection risks.
- [DATA_EXFILTRATION]: The skill describes an audit logging mechanism that writes request/response data to S3. This is identified as a legitimate compliance and monitoring feature, which includes encryption via per-tenant KMS keys.
Audit Metadata