The Agent Skills Directory

Command Execution for Validation and Cleanup: The skill instructs the agent to use various command-line tools such as az, azd, uv, and curl. These are used to initialize projects, run local servers, and perform validation checks. Additionally, it provides shell commands for cleaning up resources, such as terminating processes bound to specific ports. These actions are standard for local development and validation workflows.
External Software Installation: The instructions include installing the Azure AI Agents extension via azd and installing Python dependencies using uv or pip from a sample's requirements.txt. While these involve external downloads, they target official registries and tools associated with the vendor's platform.
Dynamic Test Script Generation: The skill describes the creation of a 'playbook'—a self-contained Python script generated to automate the local smoke test. This script uses subprocess.Popen to manage the sample's server process during validation. This represents a dynamic execution pattern intended to streamline testing within the local environment.
Local Environment Management: The skill prompts for the creation of .env files to store configuration details such as project endpoints and deployment names. These are standard practices for configuring local development environments.
Indirect Prompt Injection Surface: The skill involves processing and executing code from external 'sample' directories. As a security consideration, when an agent is instructed to run external code, the integrity of that code should be verified to ensure it does not contain unintended instructions. The skill includes guidance to help the agent correctly identify the intended entry points within these samples.

foundry-hosted-agent-validation