The Agent Skills Directory

Trusted External Resource Interaction: The skill fetches documentation from Microsoft Learn (learn.microsoft.com) to inform its resolution logic. This interaction targets a well-known and trusted service for authoritative technical information.
Authorized Command Execution: It executes local Node.js utility scripts (e.g., verify-alm-prerequisites.js, list-tenant-envs.js) and Azure CLI commands to interact with the Business App Platform (BAP). These commands are used to manage environment lifecycles and configuration within the user's tenant context.
Consent-Based Operations: The skill implements numerous 'consent gates' using the AskUserQuestion tool. It requires explicit user confirmation, including echoing of tenant display names, tenant IDs, and environment identifiers, before performing sensitive provisioning or configuration tasks.
Standard Secret Handling: It acquires OAuth access tokens via the Azure CLI (az account get-access-token) for specific Microsoft resource audiences. These tokens are used solely for authenticating against official Microsoft API endpoints like service.powerapps.com and graph.microsoft.com.
Indirect Prompt Injection Surface: The skill ingests content from external web pages (Microsoft documentation). While this introduces a potential surface for indirect prompt injection, the risk is mitigated by using highly trusted, official Microsoft sources to guide its internal decision tree.

ensure-pipelines-host