prose  by openprose

Programming language for AI sessions and agent workflows

Created 1 week ago

New!

514 stars

Top 60.9% on SourcePulse

GitHubView on GitHub
1 Expert Loves This Project
Project Summary

OpenProse is a programming language designed for orchestrating complex AI agent workflows, addressing the limitations of traditional explicit coordination code and inflexible frameworks. It targets engineers and researchers building multi-agent systems, offering a novel approach where the AI session itself acts as an intelligent Inversion of Control (IoC) container, wiring agents based on declared intent and context. This provides structured control flow with natural language flexibility, enabling powerful AI workflow automation without external dependencies.

How It Works

OpenProse defines a virtual machine (VM) specification that any "Prose Complete" system (a model + harness) can instantiate. The AI session serves as both the interpreter and runtime, acting as an IoC container that intelligently wires agents by understanding context and intent, rather than simple name matching. This "intelligent IoC" approach inverts traditional orchestration, running inside the agent session. The unique ... syntax allows breaking structured execution for direct AI judgment and semantic evaluation of prompts or conditions, enabling dynamic decision-making.

Quick Start & Requirements

  • Install: Via Claude Code plugin: /plugin marketplace add https://github.com/openprose/prose.git, then /plugin install open-prose@prose. Restart Claude Code and run /prose-boot.
  • Prerequisites: Requires a "Prose Complete" system, such as Claude Code, OpenCode + Opus, or Amp + Opus.
  • Links: Language Specification (prose.md), Examples (28 provided, e.g., 01-hello-world.prose).

Highlighted Details

  • Session as Runtime: The AI session itself functions as the interpreter, runtime, and IoC container.
  • Intelligent IoC: Agents are wired using AI's understanding of context and intent, not just static configuration.
  • Fourth Wall (...) Syntax: Enables direct AI judgment and semantic evaluation, breaking structured execution for dynamic decision-making.
  • Open Standard, Zero Lock-in: The language specification is portable across any Prose Complete system, ensuring .prose files work universally.

Maintenance & Community

The provided README does not detail specific contributors, sponsorships, or community channels like Discord/Slack. Issue reporting is directed to the project's GitHub repository.

Licensing & Compatibility

  • License: MIT License.
  • Compatibility: Permissive MIT license generally allows for commercial use and integration with closed-source projects.

Limitations & Caveats

The project is currently in Beta. Telemetry is enabled by default for anonymous usage data collection (opt-out available). Users should expect potential bugs and are advised not to use OpenProse for critical or production workflows. Users bear full responsibility for actions performed by spawned AI agents and must review programs and outputs.

Health Check
Last Commit

1 day ago

Responsiveness

Inactive

Pull Requests (30d)
3
Issues (30d)
9
Star History
520 stars in the last 8 days

Explore Similar Projects

Starred by Andrew Ng Andrew Ng(Founder of DeepLearning.AI; Cofounder of Coursera; Professor at Stanford), Thomas Wolf Thomas Wolf(Cofounder of Hugging Face), and
4 more.

ag2 by ag2ai

0.5%
4k
AgentOS for building AI agents and facilitating multi-agent cooperation
Created 1 year ago
Updated 21 hours ago
Starred by Wes McKinney Wes McKinney(Author of Pandas), Chip Huyen Chip Huyen(Author of "AI Engineering", "Designing Machine Learning Systems"), and
22 more.

autogen by microsoft

0.5%
53k
Agentic framework for multi-agent AI applications
Created 2 years ago
Updated 3 months ago
Feedback? Help us improve.