v1.6.0#

19/11/2026

Reply CMP v1.6.0 adds safer AI and automation actions, Microsoft Fabric discovery and Azure App Service plan scheduling. It also improves optimisation scoring, AI cost labels and report design.

CMP Agent#

  • Confirmation for write actions The CMP Agent now requests explicit confirmation before using tools that change resources. Read-only actions continue without approval.

  • Improved conversations Chat now provides inline follow-up suggestions, grouped repeated tool calls, fullscreen mode and improved scrolling and dark-mode contrast. Quick actions consistently appear as compact cards, including actions reopened from conversation history.

  • Agent-designed reports Design and refine reports through an agent-guided conversation, and recover interrupted design sessions. Scheduled reports can include server-rendered charts in email content.

See CMP Agent, AI-Powered Reports and FinOps Report Service.

FinOps and AI#

  • Capacity-aware optimisation scores VM and App Service scores now reflect the most constrained CPU or memory measure, preventing low utilisation in one measure from hiding a capacity bottleneck in another. The default utilisation target is now 70%, database storage remains visible but is marked Not scored, and qualifying minimum-size Azure Redis resources can score 100.

  • Shape mismatch opportunities Optimisation results now identify resources where CPU and memory demand are imbalanced and a differently proportioned size may cost less. These opportunities do not affect the optimisation score or estimated idle-capacity cost.

  • Readable Foundry charge details Azure OpenAI and Foundry charge details now use readable model, operation, context and deployment labels across charts, legends, tables and dashboard widgets. Cost legend highlighting is more responsive, while searches, filters and exports continue to use the original billing values.

See Optimize, Analyze and Monitor.

Automation#

  • App Service plan scheduling Start and Stop policies can now manage Azure Premium v4 and Premium memory-optimised v4 App Service plans. Plans with automatic scaling or unsupported SKUs are skipped; accepted operations run asynchronously and affect every app and deployment slot in the plan.

  • Policy approval gates Automation policies can now require manual approval before execution.

  • Reliable recurring execution Recurring policies now run once per scheduled occurrence and only against resources belonging to the correct tenant. Manual runs continue to preserve the next scheduled occurrence.

See Create Automation Policies and Supported Services.

Discovery and Monitoring#

  • Microsoft Fabric capacity discovery Microsoft Fabric capacities now have full Discovery support in the Analytics category. Resource details can include state, SKU, capacity units, administrators, eligible sizes, nearby sizes, overage settings and regional capacity quota information when available.

  • Preserved Discovery filters Returning from a resource details page now restores the original Discovery category, filters and view. This also applies after moving between resources in the relationship graph.

See Browse Your Cloud Resources and Supported Services.