A high-performance orchestration skill for spawning, coordinating, and sharing state across multiple AI subagents.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install openclaw-swarm
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 openclaw-swarm 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 OpenClaw Swarm Orchestration?
OpenClaw Swarm is a specialized extension designed for developers who need to manage complex, multi-agent workflows. It provides a robust framework for spawning subagents that can share context, maintain a synchronized state, and report progress automatically. By integrating this into your Openclaw Skills library, you can move beyond single-agent limitations to build sophisticated swarms capable of solving multi-faceted problems in parallel.
This skill is built upon a dedicated fork of the OpenClaw architecture, focusing specifically on subagent lifecycle management. It enables seamless communication between parent and child agents, ensuring that every sub-task has the necessary historical context to perform accurately while minimizing redundant processing.
OpenClaw Swarm Orchestration Use Cases
- Spawning subagents with granular context sharing options such as recent messages, summaries, or full history.
- Coordinating multiple agents to work on a single project through a shared namespace in the context store.
- Executing parallel research tasks where multiple agents scan different sources simultaneously to save time.
- Building complex data processing pipelines where subagents pass state to one another via event-driven notifications.
How OpenClaw Swarm Orchestration Works
- Initialize a subagent using the sessions_spawn tool, defining the level of context to be shared from the parent session.
- Utilize the context_store to persist data in specific namespaces, allowing different agents to read and write shared variables independently.
- Trigger parallel_spawn to launch multiple tasks concurrently, selecting a wait strategy like all or race based on the desired execution flow.
- Use context_publish to send event-driven notifications to the orchestrator when specific milestones or task completions are reached.
- Monitor execution through automatic completion announcements that provide runtime data and results without the need for manual polling.
OpenClaw Swarm Orchestration Setup
To utilize these Openclaw Skills, you must be running the specific swarm-enhanced fork of the platform. Follow these steps to set up the environment:
git clone https://github.com/Heldinhow/openclaw-swarm
cd openclaw-swarm
npm install
Ensure you review the SWARM.md documentation within the repository for specific environment variables and advanced configuration options.
OpenClaw Swarm Orchestration Data Schema & Taxonomy
The skill manages data through a structured context store and session metadata to ensure consistency across the swarm.
| Attribute | Type | Description |
|---|---|---|
| contextSharing | String | Determines the depth of shared history (none, summary, recent, or full). |
| namespace | String | A unique identifier for grouping shared data in the context store. |
| wait_strategy | String | Logic for parallel tasks: all (wait for all), any (return on first, others continue), or race (return on first). |
| eventType | String | Metadata used in context_publish to categorize agent notifications. |
OpenClaw Swarm Orchestration Advanced Features
- Context Pipeline: Seamlessly pass recent chat history to subagents to maintain conversation continuity without manual prompting.
- Multi-Agent Synchronization: Use the subscribe and broadcast actions within the context store to keep multiple agents in sync.
- Flexible Wait Strategies: Optimize resource usage by racing subagents or continuing as soon as any single agent provides a valid result.
- Automated Lifecycle Reporting: Benefit from auto-announcements that detail the task result and precise runtime for every spawned subagent.
SKILL.md
Loading
Related Openclaw Skills
Freeride OpenCode
A sophisticated configuration skill for Openclaw Skills that optimizes the use of free AI models through intelligent fallback routing and provider diversification.
Feishu Automation
A powerful automation toolkit for the Feishu and Lark suite designed to handle batch document processing, data synchronization, and wiki management.
Tweet Monitor Pro
A professional scraping engine for X/Twitter and Chinese social platforms that requires no login or official API keys.
Feishu Messaging Configuration
A specialized configuration guide for optimizing how AI agents format and deliver messages within Feishu and Lark environments.
Speckit Swarm
An advanced agent orchestration system that enables parallel execution and task decomposition through specialized AI personas and ultrawork detection.
Super Browser Automation
A comprehensive framework that merges eight top-rated browser automation tools into a single, unified interface for AI-driven web tasks.
Featured*
