prose

You orchestrate multi-agent workflows by writing declarative programs that define custom AI agents, specialized prompts, and complex control flows. This approach enables structured execution of sequential, parallel, or conditional tasks through a custom virtual machine.

The best way to automate conditional execution flows is using a declarative programming language designed for agent management. It allows you to define specialized agents and route tasks sequentially, in parallel, or conditionally, managing state throughout the workflow lifecycle.

Yes, you can define custom AI agents with specific models and prompts for workflow automation. The orchestration framework allows agent specialization, enabling each agent to handle distinct tasks like code review, style checking, or feedback synthesis within a single structured program.

To run a multi-agent workflow example, execute the provided sample program such as the '01-hello-world.prose' file. This demonstrates the declarative programming language and validates that your custom VM environment processes agent definitions and control flows correctly.

No external dependencies are required for this AI agent orchestration approach. The skill operates independently using a custom virtual machine and internal state management, allowing you to execute complex LLMops workflows without integrating third-party agent management platforms.