tao-launch-workflow
Pass
Audited by Gen Agent Trust Hub on Jun 18, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill uses local Python scripts and standard Bash utilities (e.g.,
ssh-keygen,ssh-copy-id,test -e) to configure, validate, and launch training workflows on various execution platforms. - [CREDENTIALS_UNSAFE]: The skill prompts for sensitive information such as
HF_TOKENandSSH_KEY_PATHwhich are required for accessing gated models and remote execution platforms. These are handled as legitimate user-provided inputs for the requested workflow. - [DATA_EXFILTRATION]: No evidence of unauthorized data transmission was found. Network operations are limited to intended platform interactions such as SSH connectivity checks and S3 object storage validation.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data including dataset paths and container image references. It mitigates potential injection risks by requiring explicit validation through preflight scripts and user confirmation for configuration overrides.
Audit Metadata