prose

What problem does it solve?

OpenProse provides a self-contained, VM-driven language to orchestrate AI agents, sessions, and workflows, turning ad-hoc prompts into reliable, repeatable automation.

Core Features & Use Cases

• Structured orchestration: Define agents, sessions, blocks, and control flow to coordinate multi-step AI workflows.
• Parallel and sequential execution: Seamlessly run tasks in parallel or in sequence with clear composition and context passing.
• Persistent memory and state: Support for persistent agents and various state backends (filesystem, SQLite, PostgreSQL) to enable resumption and long-running tasks.
• Modular program composition: Use imports, blocks, and pipelines to build scalable, reusable automation.
• Subagent spawning via Task tool: Spawn real subagents to perform defined tasks and return binding pointers for downstream steps.

Real-world example: Coordinate a team of specialized agents (researcher, reviewer, writer) to gather, analyze, and synthesize insights, then produce a final report with traceable bindings and memory.

Quick Start

Install Crocbot and run a sample OpenProse program with prose run examples/01-hello-world.prose.