Activity
Live event feed of governance, operations, insights, and system events
Last updated 8 Oct 2026Monitor · article 1 of 2
The Activity stream is your near-real-time view of what the platform is doing. It records governance decisions, operational events, insights, and system events so you can see and investigate what happened, when, and why.
The Activity stream is the default tab of the Audit log page. Open Audit log in the sidebar, under the Governance group (or go to /workspace/audit), and you land on the Activity tab. The page's other tabs are Decisions (the policy-decision log — see Audit log), Runs, and, for admins and owners, Changes.
The Activity Feed
The feed shows recent platform events newest-first, updated in near real-time as new events arrive. Events are grouped into four categories:
- Governance — guardrail evaluations, approvals, rejections, and policy decisions
- Operations — automation and execution events, change sets, escalations
- Insights — new knowledge, calibration, and analysis events
- System — connection errors, job failures, and other infrastructure events
Each event shows a timestamp, category, priority, and context. The feed loads a page at a time (20 events) with a control to load older events.
Filtering the Feed
The Activity tab supports filtering by:
- Category — governance, operations, insights, or system
- Priority — all, actionable, or high
- Free-text search — match on event content
- Date range — choose the lookback window (1–365 days, default 30) via the date selector in the page header
You can also mark individual events as read, or mark all visible events as read, to keep the feed focused on what's new.
Exporting
Use the Export button in the page header to download the filtered activity feed as a CSV (activity.csv). The export honors the active category, priority, search, and date-range filters.
Health Indicator
Governance status sits in the top bar on every page, next to Pause agent. It reports only the conditions that stop the agent, and it is kept apart from the pending-approval count so that a waiting approval can never hide an incident:
| Indicator | What it means |
|---|---|
| Incident mode active (red) | Incident mode, the workspace-wide stop, is on. Select it to open the Queue, which shows the reason and a control to deactivate it |
| N pauses active | One or more pauses (emergency stops) are in effect. Select it to manage them under Policies > Pauses |
| Status unavailable | The status couldn't be read. Reload to check for incidents and pauses — a failed read is never shown as an all-clear |
| Nothing shown | No incident and no active pauses |
When incident mode is on, it takes the place of the pause count. The status is read from live governance state and refreshed as you move between pages (at most every 30 seconds), not from the most recent event in the feed. Pending approvals are counted separately, as a badge on the Queue row in the sidebar. Budget use is not part of the status: each in-flight campaign's share of its budget is on the Campaigns lifecycle board, and your limits are under Policies > Spend limits.
Related articles
Still stuck? Browse the FAQ or contact us.