openai-api
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill enforces secure secret management by instructing the agent to only load credentials from environment variables and strictly forbidding the hardcoding or sharing of API keys.
- [SAFE]: Implementation guidance promotes the use of structured outputs (Pydantic) and explicit schemas, which effectively mitigates indirect prompt injection risks by validating responses against a fixed model.
- [SAFE]: The included code examples follow security best practices, such as implementing redaction in logs and using official SDKs to handle network requests securely.
- [SAFE]: Diagnostic shell commands are limited to benign local environment checks and do not involve untrusted data or unsafe parameters.
Audit Metadata