prose

You orchestrate multi-agent AI workflows by writing declarative `.prose` program files that define sequences, parallel tasks, loops, and error handling for agent execution. This approach turns AI interactions into reproducible, manageable processes with structured state persistence and context passing.

Declarative AI orchestration is the process of defining multi-agent workflows through structured program files rather than imperative code. You need it when managing agent lifecycles, context passing, and complex control flow patterns like parallel execution or conditional logic for sophisticated AI systems.

Yes, you can define reusable agent configurations with specific models, prompts, and permissions within the workflow program files. This allows you to assign distinct roles to different agents, such as having one identify bugs while another checks style guide adherence.

You run a multi-agent workflow by executing structured `.prose` program files that contain your declarative workflow definitions. You can start by running the included example program 'examples/01-hello-world.prose' to see how the language orchestrates agent sequences and context passing.

Yes, the declarative workflow language supports parallel execution, loops, conditional logic, and error handling for AI agents. These complex control flow patterns allow you to build sophisticated orchestration processes, such as automating a multi-step code review and feedback synthesis.

The best way to automate a multi-step code review is to define a declarative workflow where one agent identifies bugs, another checks style adherence, and a third synthesizes feedback into a single report. This structured program manages agent lifecycles and state persistence automatically.