Skip to main content
Aegis records the number of input and output tokens consumed by each agent action and surfaces this data in the Token Usage dashboard. You can see token consumption broken down by session, compare usage across time ranges, and track your account against the free tier monthly limits.

Where to view token usage

Open Dashboard > Token Usage in the sidebar. The page shows a stat summary strip, usage charts, and a full record table. Use the range filter, Today, 7 days, 30 days, or All time, to scope the data to the period you want to inspect.

What the dashboard shows

Summary stats

The stat strip at the top of the page shows four numbers for the selected range:

Usage by session chart

The bar chart groups input and output tokens by session_id. Each session appears as a pair of bars, input (dark) and output (orange), so you can see which sessions consumed the most tokens at a glance. Hover over any bar to see the full session ID and the exact token counts for that session.

Input vs output breakdown

A donut chart shows the proportion of input tokens to output tokens across all sessions in the selected range.

Token records table

The table at the bottom of the page lists individual token meter records. Each row corresponds to one agent action and includes:
The dashboard loads your complete token history for the selected range. If you have a large number of records, the initial load may take a moment.

Free tier limits

Aegis’s free tier includes the following monthly caps: Once you reach a cap, Aegis will stop processing new actions of that type for the remainder of the billing period.

Notification thresholds

Aegis sends you dashboard notifications as your token budget is consumed. Alerts fire at the following thresholds: You can adjust notification preferences in Settings.
If you frequently approach the 75% threshold early in the month, review your agent sessions for actions with unexpectedly high output token counts. Long model responses on complex tool calls are the most common source of token spikes.