The Agent Skills Directory

[Command Execution]: The skill instructs the agent to perform routine developer operations using local scripts, such as npm run export-policy-data to generate policy catalogs and npm run mock-policy-server for local verification. These commands are intended to be run within the user's controlled development environment.
[Configuration and Policy Management]: The agent is guided to modify project configuration files, including policyData.jsonc and product.json. It also describes the architecture for how the system interacts with platform-specific policy stores like the Windows Registry and macOS managed preferences to enforce enterprise settings.
[Resource Integrity]: The documentation references authoritative schemas and source code from trusted repositories (e.g., official Copilot runtime and VS Code repositories) to maintain consistency between policy declarations and runtime enforcement.
[Indirect Prompt Injection Surface]: As a tool for creating and modifying policies based on user input, the skill handles data that could potentially contain malicious instructions. To address this, the guidelines mandate the use of automated export and validation tools that detect metadata conflicts and verify the integrity of the policy catalog before changes are finalized.

policy-and-managed-settings