Vercel's Agent Runs now surfaces subagent activity for eve projects, giving developers a direct view into what each subagent did, when, and at what cost.
The new Subagents tab organizes every subagent by the turn that spawned it. Each row exposes the prompt, duration, and failures on a shared timeline. Click into any subagent and you get the full run detail: turns, tool calls, metadata, token usage, and cost, identical to what a parent run shows.
The value here is not just visibility. It is accountability across multi-agent chains. If a subagent fails silently or burns tokens unexpectedly, this is where you find it. Check the Agent Runs dashboard or the official documentation to see how deep the trace data actually goes.
[READ ORIGINAL →]