> ## Documentation Index
> Fetch the complete documentation index at: https://platform.atlan.com/llms.txt
> Use this file to discover all available pages before exploring further.

> ## Agent Instructions
> To act on Atlan objects, use the Atlan MCP server at https://api.atlan.com/mcp or the atlanai CLI; `atlanai --map json` prints its command map. Run a read-only identity check before any write.
> The docs MCP server at /mcp searches these docs only. It cannot read or change Atlan objects.
> SDK packages: Python `atlanai` (PyPI) and TypeScript `@atlanai/sdk` (npm). Show Python first, then TypeScript.

# Agents usage

> Compare agent usage, cost, efficiency, and recorded errors across your account.

Use **Agents usage** to compare agents that produced recorded runs in the
selected period. It starts with the fleet summary, then lists the agents that
contributed to it.

<img src="https://mintcdn.com/atlan-602e2b74/hJeuncFzfbRCdu_j/assets/screenshots/composed/agent-usage.png?fit=max&auto=format&n=hJeuncFzfbRCdu_j&q=85&s=58458d6f671889a26514cfd7227facee" alt="Agents usage with a fleet summary, spend trend, and agent table" width="2400" height="1600" data-path="assets/screenshots/composed/agent-usage.png" />

## Review the fleet

1. Expand **Reporting** in the sidebar.
2. Select **Agents usage**.
3. Read **What changed** before comparing individual agents. It shows spend,
   runs, active agents, recorded span error rate, and cost per run for the
   selected period.
4. Use the trend control to switch the chart between cost, tokens, runs,
   efficiency, and recorded error rate.

The report labels a value as unavailable when its source could not be read. It
does not turn an unavailable read into a zero.

Agent Registry calculates each run's cost from the token counts on its trace
and your organization's rates. See [FinOps](/finops) for how a cost is
calculated and how to read it.

## Find an agent that needs attention

Use the agent table to sort by spend, runs, cost per run, tokens per run,
recorded error rate, or P90 latency. Search by name when you already know the
agent. Select a row to open that agent's profile and inspect its usage and
recent runs.

## Read the metrics carefully

* **Span error rate** measures recorded spans marked as errors. It does not
  measure response quality.
* **Cost per run** and **tokens per run** are period averages. Compare them
  with run volume before treating a change as a trend.
* **Runs using skills** counts runs with recorded skill use. A missing value
  means the report could not establish the breakdown.

<Note>
  **AI usage** is a separate report. **Skill usage** remains Coming soon in the
  reporting navigation; it is not an alternate view of Agents usage.
</Note>