The Agent Skills Directory

[SAFE]: The skill interacts exclusively with official TrueFoundry platform domains (truefoundry.com, truefoundry.cloud) and trusted container registries such as AWS ECR, GHCR, and NVIDIA NGC.
[SAFE]: Credential management is handled securely via environment variables and .env files. The documentation provides explicit instructions for the agent to avoid logging or exposing secret tokens.
[SAFE]: The included helper script tfy-api.sh implements security validations, including checks for allowed HTTP methods and protection against path traversal in API paths.
[SAFE]: Python package dependencies, specifically the truefoundry SDK, are vendor-specific and version-pinned.
[SAFE]: The skill demonstrates high security awareness by explicitly instructing the agent to avoid ingesting content from untrusted third-party release pages to prevent indirect prompt injection risks.

docs