operations
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided operational contexts and infrastructure descriptions to generate deployment plans containing executable commands. * Ingestion points: User requests describing deployments, CI/CD pipelines, or Kubernetes manifests. * Boundary markers: Defined output sections (risk, plan, checks, rollback, security) are used to structure the agent's response. * Capability inventory: Generates operational plans containing shell commands for infrastructure management (e.g., kubectl, git). * Sanitization: Includes a Python validation script (scripts/validate-output.py) that programmatically enforces safety requirements, such as mandatory human approval gates for production or destructive operations.
- [DYNAMIC_EXECUTION]: The repository contains a test execution script that invokes a local validation utility during the development or testing phase. * Evidence: tests/run-validation.py uses subprocess.run to execute scripts/validate-output.py with shell=False to verify output compliance. * This behavior is localized to the skill's internal validation logic and does not incorporate untrusted external input into the execution path.
Audit Metadata