The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions to install the google-adk library from its official repository on GitHub. These downloads originate from a well-known organization's infrastructure.
[DYNAMIC_EXECUTION]: The toolkit includes a code_execution tool that allows agents to perform computational tasks. This is a primary feature of the framework designed for building autonomous agents.
[INDIRECT_PROMPT_INJECTION]: The skill establishes an attack surface by enabling agents to ingest data from external sources via tools like Google Search. The documentation recommends best practices such as tool_confirmation flows to mitigate risks from processing untrusted content.
Ingestion points: Data enters the agent context through the google_search tool output described in SKILL.md.
Boundary markers: The skill suggests using tool_confirmation=True to require human approval before sensitive tool execution.
Capability inventory: The framework supports code_execution and deployment to cloud environments.
Sanitization: Standard LLM safety guardrails are assumed; specific data sanitization logic is not detailed in the provided code snippets.
[COMMAND_EXECUTION]: The skill includes standard commands for containerizing applications and deploying them to Google Cloud services using docker and the gcloud CLI.

google-adk-python