agentrun
A workflow language for defining an existing agent's work as repeatable steps and decisions.
SITE CAPTUREBased on information published on the service's website. Selection and review policy
WHAT IT DOES
What does it do?
AgentRun is a workflow language for existing agents. Workflows can be written with JSON or a TypeScript builder, and support calls between workflows, parallel tasks, and loops with a limited number of iterations. In the example, it checks an answer, its sources, and decision confidence, then sends results that fail the conditions to a review step.
KEY FEATURES
Key features
- Define repeatable work steps and call agents when needed.
- Use Jev's decisions and confidence to choose the next step.
- Write workflows with JSON or a TypeScript builder and Zod contracts.
- Run other workflows, parallel tasks, and loops with a limited number of iterations.
- Send results that fail the configured conditions to a review step.