managing-llm-configs
Pass
Audited by Gen Agent Trust Hub on Jul 15, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides an integration guide for well-known and trusted services including LaunchDarkly, Cloudflare, OpenFeature, Vercel, and PostHog. These references are consistent with the skill's stated purpose of managing remote LLM configurations.
- [SAFE]: The instructions explicitly advocate for security best practices, such as using Zod schemas to validate untrusted LLM outputs. This approach helps mitigate risks associated with code/prompt drift and ensures that the application handles non-conforming responses safely.
- [SAFE]: The skill describes a robust failure model using 'disabled', 'managed', and 'degraded' states to ensure that unavailability of remote configuration services does not break the core functionality of the application.
Audit Metadata