Summary
On July 3, 2026, Vercel made Agent Runs available in its MCP server and CLI: traces from agents deployed to Vercel (starting with the open-source 'eve' agent framework) are ingested automatically and can be listed and inspected, including reasoning, tool calls, token usage, and subagent data, via new MCP tools and CLI commands.
What changed
Vercel added Agent Runs to the Vercel MCP server and CLI. Traces from agents deployed on Vercel (initially the open-source eve framework) are auto-ingested; new MCP tools (list_agent_run_projects, list_agent_runs, get_agent_run, get_agent_run_trace) and CLI commands with --json output let developers and agents find projects with runs, list runs, and retrieve full traces including reasoning, tool calls, token usage, and subagent data.
Why it matters
Exposing agent runs over MCP and the CLI makes Vercel's agent execution observable and scriptable from the surfaces developers already automate, and because traces render as markdown when piped, coding agents without MCP access can call the CLI to debug their own runs, tightening the loop between building, deploying, and supervising agents on Vercel.
Evidence excerpt
"Agent Runs now available in the Vercel MCP and CLI": eve traces are auto-ingested when deployed to Vercel; new MCP tools and CLI commands list runs and retrieve full traces including reasoning, tool calls, and token usage (Vercel Changelog, July 3, 2026).