OpenProse is an AI-native programming language that turns LLMs into virtual machines capable of orchestrating complex multi-agent workflows.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install prose
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install prose using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is OpenProse?
OpenProse is a specialized programming language designed specifically for AI sessions. It treats Large Language Models as simulators; by providing a detailed system description, the LLM doesn't just describe a task but simulates a virtual machine (VM) to execute it. This approach, where simulation with sufficient fidelity becomes implementation, allows developers to build robust, reproducible AI workflows. As a core component of Openclaw Skills, OpenProse provides the framework for agents to manage memory, state, and tool calls through a structured .prose specification.
The skill enables users to define programs that the AI agent interprets and runs. Whether you are managing simple logic or complex, multi-step pipelines involving several sub-agents, OpenProse provides the necessary runtime environment. It maps task tools to session spawning and file I/O to read/write operations, ensuring that the AI operates within a well-defined execution context.
OpenProse Use Cases
- Orchestrating multiple AI agents to work together on a single project script.
- Creating reusable and reproducible AI workflows for code reviews, research, or content generation.
- Running simulation-based programs where the LLM acts as a dedicated virtual machine.
- Executing remote AI programs directly from a URL or a centralized registry.
- Building complex automated pipelines with built-in error handling and retry logic.
How OpenProse Works
- The user triggers the skill by using any prose command or referencing a .prose file.
- The system resolves the program location, whether it is a local file, a remote URL, or a registry shorthand (e.g., handle/slug).
- The skill loads the OpenProse VM semantics from prose.md along with the appropriate state backend such as filesystem, context, or database.
- The AI agent embodies the VM, using its conversation history as memory and its tools as instructions.
- The program is executed statement-by-statement, spawning sub-sessions or performing I/O as defined in the script.
- Execution progress is tracked using a specialized narration protocol to ensure transparency and success validation.
OpenProse Setup
To begin using OpenProse within Openclaw Skills, ensure your workspace is initialized. You can set up your configuration in a .env file within the .prose directory:
mkdir -p .prose/runs .prose/agents
echo "OPENPROSE_TELEMETRY=enabled" > .prose/.env
To verify your installation and see available commands, run:
prose help
If you are migrating from an older version, use the update command to automatically restructure your workspace:
prose update
OpenProse Data Schema & Taxonomy
OpenProse organizes data across local project directories and user-level global storage:
| Directory | Purpose | Scope |
|---|---|---|
| .prose/.env | Configuration and environment variables | Project |
| .prose/runs/ | Runtime state and execution logs | Project |
| .prose/agents/ | Project-scoped persistent agent definitions | Project |
| ~/.prose/agents/ | User-scoped persistent agents | Global |
State can be managed through various backends, including standard Markdown files, In-context memory, SQLite, or PostgreSQL for high-concurrency team environments.
OpenProse Advanced Features
- Multi-Agent Spawning: Use the session statement to dynamically trigger sub-agents for specialized tasks.
- Pluggable State Backends: Switch between filesystem, SQLite, or PostgreSQL for persistent memory management.
- Remote Program Resolution: Execute scripts directly from GitHub or the OpenProse registry without local installation.
- Compiler Validation: Use the built-in compiler to validate .prose files for syntax errors and logic flaws before execution.
- Pattern Guidance: Access built-in best practices and anti-patterns directly while authoring new AI programs as part of the Openclaw Skills ecosystem.
SKILL.md
Loading
Related Openclaw Skills
Moltbook CLI
An essential CLI interface for AI agents to communicate, share updates, and engage with the Moltbook social network.
Moltscope
Moltscope provides AI agents with real-time Solana market data, autonomous trading capabilities, and a shared social layer for cross-agent intelligence.
Agent Orchestration
A comprehensive framework for implementing multi-agent coordination patterns, enabling parallel task execution and complex agentic workflows within Openclaw Skills environments.
Claw-Clawbridge
A high-signal scouting agent that runs nightly to bridge you with the right partners, clients, and collaborators.
Perstudio AI Image & Video Generation
A professional-grade AI media engine for Openclaw Skills that transforms natural language into high-fidelity images, videos, and audio.
Firecrawl Scraper
A powerful web scraping and crawling tool that utilizes the Firecrawl API to transform websites into clean, structured data optimized for AI applications.
Featured*
