Console
Describe what an agent should do, and the console proposes the harness for it.
From Start, write what the agent does for your users; the console builds the harness, its model, instructions, skills and connectors, and saves it in your workspace.
What it is
Instead of filling in a harness field by field, you describe what the agent should do for the people who use it. The console builds the setup: the agent harness, the model, instructions written for your users, the skills (up to three new ones of its own), and the connectors and apps it needs. Anything it needs that you have not connected yet is listed, and works once connected.
Using it
- In the console, open Start and describe the job in a sentence or two.
- When it is done, it saves the new harness in your workspace, with an example request for a first test.
- Review and change anything before you run a task.
What it does not do
- It runs nothing. The first task starts when you start one.
- A build that fails saves nothing.
- Configured agent / harnessEvery part of a harness and what it controls.
Ready to run this against a live workspace?Create API key