tao-launch-workflow
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes local Python helper scripts from the TAO skill bank (such as list_tao_platforms.py and check_tao_launch_preflight.py) to validate the environment and confirm platform configuration before submitting jobs.
- [SAFE]: The instructions mandate a 'Non-Negotiable Launch Gate' that requires explicit user confirmation for the container image, dataset paths, and compute resources before any side-effecting actions occur.
- [SAFE]: Credential management is handled securely; the skill is instructed to source environment variables from specific user-approved files while strictly forbidding the printing or logging of these secrets.
- [SAFE]: Infrastructure operations, such as setting file permissions for SSH keys (chmod 600), follow established security best practices for remote cluster access.
- [SAFE]: The skill remediates missing dependencies by requesting user approval to install official vendor-provided SDKs, ensuring a controlled and trusted software supply chain.
Audit Metadata