Observability & ops
What this is for. Observability (/observability) is the telemetry surface for this instance: traces, cost, latency, anomalies, ensemble (God Mode) races, and what the model actually received. Ops (/ops) is remediation. Hands, remote nodes, extensions, and notification preferences are not on this page — they have their own chapters.
| Area | Route | Meaning |
|---|---|---|
| Observability | /observability |
Metrics / tracing UI — tabs Usage, God Mode, Context |
| Ops | /ops |
Kubernetes ops agent — observe → diagnose → propose → approve → apply. Default propose-only. Cluster URL, kubeconfig, and GitOps repo are instance config, not product defaults. |
Elsewhere (not this page): Hands (/hands), Remote nodes (/nodes) — including guarded SSH invoke, Ingress (/ingress), Extensions (/extensions), Notifications (/notifications-settings).
Usage tab
Section titled “Usage tab”Usage is token telemetry: Total Traces, Total Cost, Avg Latency, Anomalies, daily cost, model distribution, and the trace table (timestamp, model, provider, tokens, cost, latency, tools, quality).
God Mode tab
Section titled “God Mode tab”/observability has three tabs: Usage (existing traces / stats), God Mode, and Context. The God Mode tab lists ensemble runs (conversation, winner, model count, cost, duration, whether a tie was broken), win-rate by model, and the average cost multiple versus a single model. Click a run to open that conversation’s God tab (step log, who voted for whom, and each model’s comments on the others).
How a race is set up, how the winner is chosen, and how to read the conversation God tab: Conversations — God Mode.
Context tab
Section titled “Context tab”The Context tab answers a question nothing in EYAS could answer before: what the model actually received, not what was meant to be sent. It shows the average and peak token cost of each prompt section (and how many samples that rests on), truncation frequency (how often — and which section — gets cut to fit budget), and estimate vs. actual: the gap between the token estimate and what the provider reported, the first way to measure how far that estimate drifts.
Detailed per-section records are short-lived by design (7 days by default); only the daily rollup survives long-term. If you go looking for older detail and can’t find it, that’s expected, not data loss.