The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill facilitates the installation of development tools using standard package runners like uvx and npx. These commands fetch content from verified registries and official repositories to set up the agent development environment.\n- [COMMAND_EXECUTION]: The skill provides a comprehensive CLI for managing the agent lifecycle, including commands to scaffold project structures, run local simulations, lint code, and provision infrastructure on Google Cloud. These operations are essential to the tool's primary purpose and operate within the user's project context.\n- [DYNAMIC_EXECUTION]: The skill supports the creation and execution of agent logic using local Python scripts and the Agent Development Kit (ADK) library. It employs configuration files to resolve agent modules and manage deployment targets, representing standard patterns for extensible development tools.\n- [INDIRECT_PROMPT_INJECTION]: As a tool designed for building AI agents, it provides mechanisms to process user queries and evaluation datasets. While these represent ingestion points for external data, the risk is consistent with the intended functionality of an agent development platform and is managed within the development lifecycle.\n
Ingestion points: Evaluation test cases in evalset.yaml, interactive user queries via agents-cli run, and document ingestion for RAG systems.\n
Boundary markers: The skill documentation provides examples of prompt templates where users are encouraged to define instructions and context clearly.\n
Capability inventory: The tool possesses capabilities required for enterprise agent deployment, including shell execution for infrastructure management and network access for cloud service interaction.\n
Sanitization: The skill utilizes the ADK framework and standard Python libraries, providing a foundation for users to implement their own application-level validation and sanitization.

google-agents-cli