azure-cloud-migrate

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • Indirect Prompt Injection: The skill is designed to ingest and process a variety of untrusted workspace data, including application source code, Kubernetes manifests, and CloudFormation templates. While this is necessary for its migration assessment and code conversion tasks, it presents a potential attack surface where malicious patterns in the source data could influence the agent's behavior.
  • Ingestion points: Application source code (.js, .py, .java, .cs), Kubernetes manifests (.yaml), CloudFormation/SAM templates (.yaml), and platform-specific configurations (app.yaml, Procfile).
  • Boundary markers: The skill provides strict reporting templates (e.g., migration-assessment-report.md) with mandatory headings to structure output and mitigate accidental command obedience.
  • Capability inventory: The skill utilizes significant system capabilities including the Azure CLI (az), Kubernetes CLI (kubectl), Google Cloud CLI (gcloud), and Docker for resource discovery, image migration, and infrastructure deployment.
  • Sanitization: The instructions explicitly include safety measures such as auditing service discovery patterns in source code and requiring explicit user confirmation via ask_user for all destructive actions.
  • External Resource Management: The migration workflow involves interactions with remote cloud platforms and container registries. The skill utilizes official CLI tools to perform these operations and references well-known service providers (Microsoft, AWS, Google Cloud) for image migration and dependency management. These activities are consistent with the skill's purpose and follow established vendor resource patterns.
  • Sensitive Data Handling: The skill provides detailed guidance for migrating secrets to Azure Key Vault. It implements security-conscious practices, such as recommending the use of temporary files or shell pipes to prevent credentials from being logged in shell history or stored in plain text environment variables.
  • Dynamic Resource Configuration: The skill generates infrastructure-as-code (IaC) templates and Azure-compatible code based on the source environment. This represents a complex operation where the agent produces executable configurations from analyzed input. Users should review all generated code and templates before applying them to their environments.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 10:52 AM