adk-deploy-guide

Pass

Audited by Gen Agent Trust Hub on Mar 5, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill fetches documentation and reference implementation guides from Google's official GitHub pages (google.github.io), which is a trusted source.
  • [COMMAND_EXECUTION]: Instructions include the use of the adk CLI, terraform, and make for managing infrastructure and deploying agents. These commands are standard for the intended devops and deployment workflows.
  • [REMOTE_CODE_EXECUTION]: The guide recommends using uvx to execute the agent-starter-pack utility. This tool is part of the ADK ecosystem used for project scaffolding and CI/CD setup.
  • [CREDENTIALS_UNSAFE]: The skill demonstrates secure secret management by providing instructions on using Google Cloud Secret Manager, specifically advising against passing sensitive keys as environment variables.
  • [PROMPT_INJECTION]: Reference patterns for event-driven agents in references/event-driven.md process data from external triggers like Pub/Sub and Eventarc. While this introduces an indirect prompt injection surface common to all autonomous agents, the documentation provides the standard implementation pattern for these services.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 5, 2026, 06:36 PM