SwarmClaw: Self-Hosted AI Agent Runtime for Autonomous Swarms

Introduction

SwarmClaw is an open-source, self-hosted AI agent runtime and multi-agent framework designed for autonomous agent swarms. It provides a robust platform for running AI agents, orchestrators, and agent swarms with features like durable agent memory, Model Context Protocol (MCP) tools, schedules, and delegation. Supporting over 23 LLM providers, including Claude, GPT, Gemini, OpenRouter, and Ollama, SwarmClaw offers a practical alternative to solutions like Claude Code and LangChain for self-hosted AI workflows.

Developed in TypeScript, SwarmClaw has garnered significant community interest, boasting 645 stars and 137 forks. It is released under the MIT License, ensuring flexibility for developers and organizations.

Installation

SwarmClaw offers multiple installation methods to suit different user needs, from desktop applications for non-technical users to Docker for containerized deployments.

Desktop App (Recommended for non-technical users)

Download the one-click installer for macOS (Apple Silicon & Intel), Windows, and Linux (AppImage + .deb) from the official website. Download SwarmClaw Desktop App

Global Install (npm)

For developers, SwarmClaw can be installed globally via npm, yarn, pnpm, or bun.

npm i -g @swarmclawai/swarmclaw
swarmclaw

Running swarmclaw starts the server on http://localhost:3456.

From the Repository

To run SwarmClaw directly from its source code:

git clone https://github.com/swarmclawai/swarmclaw.git
cd swarmclaw
nvm use
npm run quickstart

npm run quickstart handles dependency installation, local configuration, and starts SwarmClaw.

Docker

For a containerized setup, use Docker Compose:

git clone https://github.com/swarmclawai/swarmclaw.git
cd swarmclaw
mkdir -p data
touch .env.local
docker compose up -d --build

Access SwarmClaw at http://localhost:3456 after deployment.

Examples

SwarmClaw is a versatile agent runtime, enabling a wide range of use cases for both individuals and teams.

Personal Assistant

Deploy a single agent with memory, web access, scheduling, and file tools to act as your always-available copilot. It can remember preferences, schedule reminders, research, draft content, and manage your daily tasks.

Virtual Company

Build a full organizational chart of specialized agents that collaborate, delegate, and report up, simulating a real company. Each agent can have its own provider, model, personality, and tool access, with a CEO agent delegating tasks to department heads.

Development Team

Create a squad of agents mirroring an engineering team, capable of planning, building, reviewing, and testing in parallel. A Lead agent can break down work, Dev agents can use coding tools like Claude Code, QA agents can run tests, and Reviewer agents can audit pull requests.

Research Bureau

Spawn a swarm of research agents to work in parallel, each with different search strategies, and then synthesize their findings. Agents can search, fetch, read, and summarize independently, with a lead agent collecting outputs into structured reports.

Why Use SwarmClaw?

SwarmClaw distinguishes itself by focusing on several core capabilities crucial for advanced AI agent deployments:

• Delegation, Orchestration, and Background Execution: Facilitates complex workflows with delegated work, orchestrator agents, subagents, durable jobs, and background task execution.
• Structured Sessions: Offers reusable, bounded runs with templates, facilitators, participants, and durable transcripts, ideal for structured collaboration and problem-solving.
• Autonomy and Memory: Features heartbeats, orchestrator wake cycles, schedules, long-running execution, durable memory, reflection memory, and project-aware context for truly autonomous agents.
• Extensive LLM Provider Support: Connects to 24+ built-in providers, including major cloud LLMs and local options like Ollama, giving users flexibility and control over their models.
• Self-Hosted Control: Provides a self-hosted environment, ensuring data privacy and full control over your AI agent infrastructure, a key advantage over cloud-dependent solutions.
• Operator Tooling: Includes connectors for various platforms (Discord, Slack, Telegram), extensions, browser automation, shell/files/git tooling, and runtime guardrails for comprehensive management.

Links

• GitHub Repository: https://github.com/swarmclawai/swarmclaw
• Official Documentation: https://swarmclaw.ai/docs
• Website: https://swarmclaw.ai
• Discord Community: https://discord.gg/sbEavS8cPV
• SwarmDock Marketplace: https://swarmdock.ai
• SwarmFeed Social Network: https://swarmfeed.ai