The Agent Skills Directory

[SAFE]: The skill manages authentication securely using environment variables (TFY_API_KEY, TFY_BASE_URL) or local .env files. The tfy-api.sh helper script uses a safe line-by-line parser for configuration files to avoid vulnerabilities associated with shell sourcing.
[SAFE]: Network communications are directed to the user's platform URL or vendor-owned domains. The API helper script includes validation logic to ensure that HTTP methods are restricted to a whitelist and that API paths do not contain traversal sequences (..).
[SAFE]: External dependencies, such as the truefoundry CLI and container images (vLLM, TGI, NVIDIA NIM), originate from official vendor repositories or well-known, trusted registries including AWS ECR, GitHub Container Registry, and NVIDIA NGC.
[SAFE]: The skill enforces operational safety by requiring the AI agent to obtain explicit user confirmation before selecting or deploying to a workspace, mitigating the risk of unintended modifications to the user's environment.

ml-repos