← All articlesAI INSIGHT

MACHINE INTELLIGENCE

Claude Code regularly sends each background agent’s transcript to its model

How much is that status line costing you?

A long transcript passes through a terminal into a tiny blank status chip while counters accumulate below, illustrating the token cost of background-agent status summaries.

How much is that status line costing you?

According to the GitHub issue report, Claude Code sends each background agent's full transcript to that agent's model about every 30 seconds to generate a 3 to 5 word status label. That label updates the agents panel; it never enters the conversation.

The GitHub issue report identifies the associated cost in OpenTelemetry traces under agent_summary. OpenTelemetry is the telemetry export used to inspect those requests.

One important distinction: the spending figures are Claude Code's own per-request estimates, not an invoice.

For builders running background agents, the practical next step is to check their own OpenTelemetry traces for agent_summary and assess the estimated cost in their own workload.

Claude Code regularly sends each background agent’s transcript to its model