Feature — Spend Observability

Tokenistt gives you real-time cost visibility across every LLM call in your infrastructure — by workspace, model, function, and engineer.

What you can see

• –Token throughput per minute / hour / day
• –Cost per workspace, model, and function call
• –Spend velocity vs. projection (budget burn rate)
• –Per-engineer attribution in team workspaces
• –Anomaly alerts when spend spikes unexpectedly
• –Historical trends and 30-day rolling averages

Spend anomaly detection

Tokenistt maintains a rolling baseline of spend per workspace. When any 5-minute window exceeds 2× the baseline, an anomaly event fires — visible in the dashboard and optionally sent to a webhook.

# Example anomaly event payload
{
  "event": "spend_anomaly",
  "workspace": "agents-platform",
  "window_usd": 0.84,
  "baseline_usd": 0.38,
  "ratio": 2.21,
  "ts": "2026-06-03T14:02:11Z"
}