The Agent Skills Directory

[SAFE]: The skill serves as a detailed guide for the Modal platform. It contains no malicious code, obfuscation, or unauthorized data access patterns.
[COMMAND_EXECUTION]: Examples include legitimate use of shell commands via subprocess for tasks like multi-GPU training and web serving. The documentation includes a proactive security warning in references/web-endpoints.md regarding the risks of shell=True and recommends safer alternatives using argument lists to prevent command injection.
[EXTERNAL_DOWNLOADS]: The skill references downloads and configurations from well-known sources, including official GitHub repositories (e.g., modal-labs), the Nvidia container registry, and the Hugging Face model hub. These are documented neutrally as part of the setup process.
[CREDENTIALS_UNSAFE]: The skill promotes secure secret management using the platform's native modal.Secret feature. All code examples use placeholders (e.g., KEY=value) or environment variable lookups, adhering to security best practices for handling sensitive information.

alterlab-modal