The Agent Skills Directory

External Package Installation: The skill uses pip to install official libraries such as google-genai, google-cloud-aiplatform, and openai from standard package registries. These dependencies are required for communicating with the specified cloud services.- Command Execution: The instructions utilize the gcloud CLI for authentication, API enablement, and resource discovery. Additionally, a shell script (scripts/verify_all.sh) is provided to execute the included Python sample scripts for verification purposes.- Indirect Prompt Injection Surface: The skill processes user-supplied data, including prompt text and project identifiers, to perform model inference. While the skill mandates a 'Tier R' confirmation protocol requiring explicit user approval before execution, the ingestion of untrusted content into model prompts is an inherent security consideration.- Dynamic Execution: The skill involves the automated execution of locally provided Python scripts through a verification shell script. This is used to ensure the environment is correctly configured before the agent performs inference tasks.

agent-platform-inference