prose

Execute .prose programs with a VM-like runtime for multi-agent workflows.

Updated Mar 22, 2026
One-click install
npx skills add https://github.com/MaxSaiets/myOpenCL --skill prose-maxsaiets
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: prose
Source: https://github.com/MaxSaiets/myOpenCL/tree/main/openclaw/extensions/open-prose/skills/prose
Command: npx skills add https://github.com/MaxSaiets/myOpenCL --skill prose-maxsaiets

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

OpenProse provides a compact, scriptable runtime for OpenProse AI orchestration tasks, enabling teams to design, run, and manage multi-agent workflows as structured programs. It reduces ad-hoc prompt chaining, simplifies debugging, and makes large automations reproducible by codifying agent templates, flows, and state management.

Core Features & Use Cases

  • Orchestrate concurrent and sequential agent work using sessions, parallel blocks, and pipelines.
  • Define reusable agent templates with models, prompts, skills, and permissions; compose tasks with inputs/outputs.
  • Import external programs, route inputs/outputs via a registry, and compose nested workflows with nested imports.
  • Manage state with filesystem, in-context, SQLite, or PostgreSQL backends, enabling resumption, audit trails, and cross-project reuse.

Quick Start

Write a simple .prose program and run it with prose run to see the OpenProse VM orchestration in action.

Frequently Asked Questions about prose

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I orchestrate multi-agent workflows without ad-hoc prompt chaining?

Multi-agent orchestration is handled by codifying agent templates, flows, and state management into structured .prose programs. This approach replaces ad-hoc prompt chaining, simplifies debugging, and makes large automations reproducible by executing commands through a VM-like runtime.

Can I manage persistent state for AI agent workflows using SQLite or PostgreSQL?

Yes, AI agent workflows support multiple state-backends including SQLite and PostgreSQL. You can configure these databases to manage state, enabling workflow resumption, audit trails, and cross-project reuse alongside filesystem and in-context memory options.

How do I run concurrent and sequential agent tasks in a structured program?

You can run concurrent and sequential agent tasks by writing a .prose program that utilizes sessions, parallel blocks, and pipelines. Execute the structured program using the 'prose run' command to see the OpenProse VM orchestration in action.

What is the best way to import external programs and route inputs in a multi-agent pipeline?

The best way to route inputs and outputs is via a modular skill registry that integrates with external program imports. This allows you to compose nested workflows and manage inputs/outputs effectively by routing commands through modular skill docs.

Do I need specific dependencies to define reusable agent templates with models and prompts?

No specific external dependencies are required to define reusable agent templates. You can define templates with models, prompts, skills, and permissions directly within your .prose programs, composing tasks with inputs and outputs natively.

Why use a VM-like runtime for AI automation tasks instead of standard scripts?

A VM-like runtime provides a compact, scriptable environment for AI automation that standard scripts lack. It reduces ad-hoc prompt chaining, simplifies debugging, and makes large automations reproducible by codifying agent templates and managing state across backends.