The Agent Skills Directory

[SAFE]: The skill instructs users on secure secret management, recommending the use of GitHub Secrets and App Platform secrets for sensitive environment variables like MODEL_ACCESS_KEY and DIGITALOCEAN_API_TOKEN instead of hardcoding them.
[SAFE]: All external references and endpoints utilize official DigitalOcean domains (do-ai.run, digitalocean.com), which are recognized as well-known service providers.
[SAFE]: The skill utilizes the official gradient-adk Python package, which is the vendor-provided library for building and deploying AI agents on the Gradient platform.
[SAFE]: There is no evidence of command injection, persistence mechanisms, or privilege escalation. The command examples provided (doctl, gradient agent) are standard for interacting with the service.
[SAFE]: The skill provides guidance on using guardrails and content filtering for AI agents, which helps mitigate risks associated with indirect prompt injection.

ai-services