Know when the engine needs attention
Check reachability, uptime history, cadence, latency, coverage, and failed runs in one place.
- 01Detect downtime
See the last heartbeat, restarts, and observed outage gaps.
- 02Check cadence
Confirm pulse and deep research are running on schedule.
- 03Audit execution
Review recent runs, failures, latency, and model spend.
The engine is a
Live counts, cadence, and latency from the store behind this page—not specimen metrics.
running process
Engine unreachable
No heartbeat is reaching this browser. Last observed Never observed.
No outage gap has been recorded in the observed window.
Heartbeat every 30s · restart gaps persist in the engine store · this is single-host availability, not HA
Cadence state appears once the engine is running: a pulse every 5 hours for news and insights, a deep pass every 24 hours for scale and depth.
No runs yet. New discovery, extraction, and synthesis passes appear here live.
The tiers are provider-agnostic — one base URL, one key, three model ids in backend/.env. Until a key is present the engine stores sources and serves the graph, and the model stages return a 503 that says exactly which key is missing rather than failing quietly.