google-adk
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill serves as educational documentation for the Google Agent Development Kit (ADK). No malicious code, obfuscation, or injection patterns were detected.
- [INDIRECT_PROMPT_INJECTION]: While the skill describes agents that ingest user data, it explicitly promotes security best practices to mitigate injection risks. It recommends validating tool arguments before performing side effects, checking permissions within Python tool code rather than relying on prompt instructions, and using Pydantic schemas for structured output.
- [EXTERNAL_DOWNLOADS]: The skill references official Google services such as Vertex AI Agent Engine, Cloud Run, and Google Cloud Storage (GCS). These are well-known services provided by a trusted organization.
- [COMMAND_EXECUTION]: The documentation mentions the ADK 'BuiltInCodeExecutor' tool but provides appropriate warnings regarding its usage constraints and delegation to sub-agents.
Audit Metadata