The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines prompt templates and agent structures that ingest untrusted user input, creating a potential surface for indirect prompt injection.
Ingestion points: The COORDINATOR_PROMPT in resources/prompt_template.py is designed to analyze and act upon the user's request.
Boundary markers: The prompt templates do not implement specific delimiters or instructions to ignore embedded commands within user-provided data.
Capability inventory: The LlmAgent definition in resources/agent_template.py includes a tools parameter, which allows the agent to execute delegated tasks via specialized sub-agents or external tools.
Sanitization: There is no evidence of input validation or sanitization within the provided Python templates.
[EXTERNAL_DOWNLOADS]: The skill references official Google GitHub repositories and documentation sites. These links provide legitimate resources for the Agent Development Kit (ADK) and academic research samples.

google-adk-agent