In Microsoft Sentinel, playbooks (Logic Apps) that are connected to Sentinel are most commonly run in context of an incident. From the Incidents blade, you select an incident, then choose Actions # Run playbook to trigger a manual test against that specific incident's entities and alert context. This is the recommended way to validate playbook inputs (entities, alert details, incident properties) and permissions end-to-end without changing analytics rules. While the Playbooks blade shows the Logic Apps and their connections, the incident view is where Sentinel exposes manual execution with full security operations context (assignments, comments, evidence), which is what "test a playbook manually in the Azure portal (from Sentinel)" refers to.