Unified Activity Data, Markdown in AI Summary, and Smarter PM Integrations
The update introduces fully aligned Activity Timeline views, clean Markdown rendering in AI Summary, improved Notification Service logic, and accurate channel naming with better HTML formatting in PM Agent reports.
Release dates: 01/06/2026, 01/08/2026
AI Activity improvements
Unified tooltip display for events in Week and Month view of Activity Timeline. Standardized tooltip format when hovering over Task events across different view modes. Both Week view and Month view now display a scrollable list of all event entries instead of different formats (list for Week, single entry for Month). Added tooltip display delay to improve UX when working with scroll, and updated the tooltip design for the monthly display.
Ensured complete data display across all Activity Timeline modes. Fixed the incomplete event display issue across different view modes. Week view and Month view now show the same complete set of events for identical time periods (for example, all commits and tasks for November 24 are visible in both modes). Also resolved an issue where events created seconds apart could fail to display in the monthly view.
Fixed a race condition during fast navigation in Activity Timeline. Eliminated the data mismatch issue between the displayed data and the selected period in Activity Timeline. Fixed race condition occurring during rapid week/month scrolling (multiple parallel API requests where the last response wasn't from the most recent request). Implemented a debouncing mechanism, or cancellation of previous requests, to guarantee data display only for the currently selected period.
Fixed 401 error when the token expires during Activity Timeline navigation. Resolved critical authentication error when scrolling weeks in Activity Timeline. Fixed token refresh logic so the system now correctly handles situations when the refresh token expires (after 5 minutes) while the user is working on the page. When switching time windows, all necessary data is automatically re-requested with the refreshed token instead of failing with a "Could not validate credentials" error.
PM Agent improvements
Added storage of actual chat names in PM Agent. Implemented saving of real chat names from botfarm instead of AI-generated ones. Added database migration with channel_name field, updated Thread model to store channel name, implemented channel_name retrieval from botfarm when creating threads, created script to populate names for existing records, and updated the copilot_threads VIEW and chat_agent prompts to use real channel names. PM Agent users now receive accurate chat names in context, eliminating confusion caused by AI hallucinations where Copilot invents non-existent channel names during communication analysis. Reports, activity analysis, and agent responses now use actual Slack/Telegram/Teams channel names, which is critical for accurate source identification and correct interpretation of discussion context when making decisions.
PM Agent now considers only active projects in "all projects" mode. Fixed project selection logic in PM Agent's "all projects" mode. The agent now works exclusively with active projects, ignoring deactivated ones regardless of their opened/closed status, aligning with Global Access menu logic.
Improvements and fixes
Fixed empty repository handling during Test Connection for GitLab self-hosted. Resolved critical 500 error when testing connections to GitLab self-hosted (for example, client installations in closed environments) – fixed handling of empty repositories without branches in async code that caused unhandled exceptions. Test Connection now completes correctly even when empty repositories exist in the group, and errors are no longer "swallowed" with false success reports to the frontend. Impact: Clients with self-hosted GitLab in closed environments now have working integration.