fieldrunBOOK A RUN

Your agent works here. Will it work out there?

Sample field jobs from --real-practitioner-machines

$6FIRST RUN

Install an MCP server from the quickstart

Backend engineersNode · first-time setup

Start with the published docs on your usual machine. Show us where you pause, what you try, and whether you reach a working first call.

Book this run
$9COMPATIBILITY

Run one skill inside an existing agent setup

Platform engineersWindows & Linux hosts

Use a supplied test project alongside the tools you already have installed. Capture conflicts, wrong tool calls, or silent changes in behavior.

Book this run
$8HEAD-TO-HEAD

Finish a realistic task with two tools

Staff engineersDaily agent users

Try the same defined task with each tool. Record where you switch approaches, finish by hand, or decide which you would use again.

Book this run
$7WORKFLOW

Work a normal day and mark where you drop the agent

Product engineersAgent-assisted teams

Keep your usual workflow. Flag every point where you abandon an agent run and finish by hand, and say what made you switch.

Book this run
$4DOCUMENTATION

Follow the README without asking anyone

Senior developersNew to your tool

No support channel, no search. Work only from what is written, and mark every sentence you had to read twice to act on.

Book this run
$5REGRESSION

Re-run last quarter's task on this quarter's model

ML engineersModel upgrade cycles

The same brief, the same machine, a newer model. Record what changed in the result, and whether the change was an improvement.

Book this run

Illustrative briefs only; these are not currently open participant listings.