Unified Loop & Automation, Enhanced Connector Best Practices

4 Enhanced Connector Best Practices

This release adds best-practice Skills for commonly used observability and operations connectors, making intelligent diagnostics more reliable and closer to real-world usage:

• Datadog: Improved practices for querying and diagnosing metrics, logs, and monitoring data.
• Kubernetes: Optimized recommendations for cluster resources, workloads, and event troubleshooting.
• Helm: Added best practices for release management and deployment troubleshooting.
• Docker: Improved diagnostic usage for containers and images.

Full Release Checklist

feat

• Unified loop tasks and automation into a single system, with tasks running bound to a session or independently.
• Added update time, trigger-type filtering, and time-range filtering to the search dialog, all combined together.
• Added citation sources to web search results, with one-click navigation to the original page.
• Added region-selection interaction to CSV document preview.
• Optimized the application initialization flow, now triggered uniformly via slash commands.
• Enhanced best-practice Skills for the Datadog, Kubernetes, Helm, and Docker connectors.
• On the application detail page, knowledge creation now defaults the binding target to the current application with a guiding hint.

bugfix

• Fixed an issue where the panel did not open according to artifact type (such as HTML or dashboard) when reading knowledge.
• Fixed an issue where previously read knowledge was incorrectly listed as a deliverable of the current turn.
• Fixed an intermittent failure when refreshing a dashboard.
• Fixed an issue where charts appeared blank after downloading an HTML document and opening it locally.