More reliable admin analytics and reports

This release focuses on keeping administration analytics and reporting accurate and timely, even when the system is busy publishing in the background. It also strengthens recovery after analytics delivery interruptions, so you see fewer gaps in reports and dashboards.

Stability updates for admin analytics

Administration queue counts, pipeline cards, and feed and topic reports now continue updating during continuous background publication. Queue refresh progress is also preserved when a slower card refresh needs to retry, reducing cases where refreshes appear to restart or stall.

Retained analytics, including publication and completed-job observations, can now be recovered after delivery interruptions. This helps prevent missing data in analytics views when workers finish work but previously could not deliver observations in time.

Reporting and AI usage corrections

Completed topic decisions now remain visible in reports even after related proposals are removed. This keeps reporting consistent over time, instead of dropping previously completed decisions.

Temporary AI-usage recording failures are now retried without repeating the AI request. That means you get more reliable usage tracking while avoiding accidental duplicate AI work.

Upgrade notes

If you purge source records, allow retained-observation repair to finish first. Already-removed records and overwritten older attempts cannot be reconstructed.