Suggested Training Course Outline

1. Mental model — policies vs. managed-settings vs. network vs. audit (10 min)
2. Live walkthrough: AI controls UI — policies, models, MCP tab, and the new Default availability policies (15 min)
3. Hands-on: draft a managed-settings.json with strictKnownMarketplaces, allowedMcpServers/deniedMcpServers, permissions.deny/ask/allow, and remoteControl (20 min)
4. Cloud agent specifics — firewall limitations, workflow approval gate (10 min)
5. Content exclusion + public code matching, and the IDE agent-mode gap (10 min)
6. Network layer — allowlist reference, subscription routing, IP allow list caveat (10 min)
7. Audit & metrics — what’s visible vs. what needs custom telemetry; session data sync/sharing/retention policy (10 min)
8. Q&A / apply to your org’s actual managed-settings.json (15 min)