The Agent Skills Directory

[COMMAND_EXECUTION]: Orchestrates the deployment process by executing local repository scripts, including prepare_hf_deployment.sh, manage_hf_collection.py, and verify_private_spaces.py.
[EXTERNAL_DOWNLOADS]: References the vendor's official GitHub repository (github.com/meta-pytorch/OpenEnv.git) and container registry (ghcr.io/meta-pytorch/openenv-base) for environment dependencies and base images.
[SAFE]: Dynamically modifies Dockerfiles and pyproject.toml files during the staging process to ensure correct versioning and dependency resolution.
[DATA_EXFILTRATION]: Performs network requests to Hugging Face Space domains to verify runtime functionality and health status after deployment.
[SAFE]: Evaluated for indirect prompt injection attack surfaces.
Ingestion points: Environment configuration files, local scripts, and responses from network health probes.
Boundary markers: Not explicitly implemented in the instructions.
Capability inventory: Shell command execution for staging and deployment, and network access for health verification.
Sanitization: Not observed in the descriptive workflow.

release