managing-deployment-rollbacks

Pass

Audited by Gen Agent Trust Hub on Jul 21, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides standard automation instructions and templates for DevOps tasks. No prompt injection attempts or safety bypasses were identified in the instructions.
  • [SAFE]: No hardcoded credentials or sensitive data exfiltration patterns were detected. Configuration files in the assets/ directory correctly use placeholders (e.g., YOUR_APPLICATION_NAME, YOUR_SLACK_WEBHOOK_URL) for environment-specific information.
  • [SAFE]: The provided Python scripts (rollback_deployment.py, verify_deployment.py, get_previous_version.py) are focused on local file system management, such as validating directories, copying files, and generating JSON metadata reports. They do not perform unauthorized network operations or download external code.
  • [SAFE]: The allowed-tools configuration in SKILL.md appropriately limits the agent's shell access to specific commands (git, docker, kubectl) required for the skill's stated purpose of managing deployments.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 21, 2026, 11:31 AM
Security Audit — agent-trust-hub — managing-deployment-rollbacks