What problem does it solve? Platform teams need agents that can operate an Azure platform day-to-day without giving up human decision authority or losing state during long-running approvals. This Skill designs durable agentic workflows on Dapr Workflow for AKS so provisioning approvals, health monitors, multi-step provisioning, and auto-remediation survive pauses and failures while every agent action lands as a PR or approval request. ## Core Features & Use Cases - Four workflow patterns: human-in-the-loop external-event approvals, monitor-pattern recurring checks via continue-as-new, saga/compensation for multi-step provisioning, and a Cluster-Doctor remediation loop (detect, diagnose via MCP, remediation PR, human approval). - Runtime guidance: targets Dapr Workflow on AKS, explains why Azure Container Apps managed Dapr is unsuitable for durable workflows, and when to step up to Dapr Agents for autonomous planning. - Governed control surface: generates Dapr Workflow definitions (Python or .NET), component config, GitOps/issue wiring, and exposes approval and status operations through platform API endpoints. - Use Case: A platform engineer wants failed AKS GitOps reconciliations to open an issue, have an agent diagnose it through AKS MCP tools, and propose a remediation PR that a human approves before GitOps applies it. ## Quick Start Ask the agent to design a Dapr Workflow on AKS that pauses a subscription provisioning request for human approval and compensates cleanly if any step fails.
