DISCOVER THE FUTURE OF AI AGENTSarrow_forward

HiClaw

calendar_todayAdded Apr 22, 2026
categoryAgent & Tooling
codeOpen Source
Workflow AutomationDockerMulti-Agent SystemModel Context ProtocolAI AgentsAgent & ToolingAutomation, Workflow & RPAProtocol, API & IntegrationSecurity & Privacy

An open-source collaborative multi-agent runtime platform featuring Manager-Workers architecture and Matrix protocol for transparent, secure, human-in-the-loop multi-agent task orchestration.

HiClaw is an open-source runtime platform designed for multi-agent collaboration, built around a Manager-Workers architecture where a Manager Agent handles task decomposition and orchestration while multiple Worker Agents execute specific tasks. All communication is based on the Matrix instant messaging protocol, providing inherent transparency and auditability. The platform enables Human-in-the-Loop by default, allowing humans to observe and intervene in agent behavior at any time through Matrix rooms.

On the security front, HiClaw adopts an enterprise-grade credential isolation model — Workers hold only consumer tokens, while actual API keys, GitHub PATs, and other sensitive credentials are centrally managed and proxied by Higress AI Gateway. Even if a Worker is compromised, no real credentials are exposed. MCP Server integration (v1.0.6+) is also secured through Gateway proxying.

The platform offers pluggable agent runtimes (OpenClaw, CoPaw, etc.), MinIO-based centralized shared file storage keeping Workers stateless, and on-demand skill pulling from the skills.sh community ecosystem (80,000+ skills). Deployment supports one-click Docker installation across all platforms (macOS/Linux/Windows) and Kubernetes Helm Chart declarative deployment (v1.0.9+, with custom controller hiclaw-controller, Worker Template Marketplace, Nacos Skills Registry), with a minimum requirement of 2 CPU + 4 GB RAM for a complete system.

Typical use cases include OPOC (One Person One Company — individuals managing full-stack development agent teams via a Manager), enterprise digital workforce deployment, and secure, compliant multi-agent collaboration environments.

Installation

macOS / Linux:

bash <(curl -sSL https://higress.ai/hiclaw/install.sh)

Windows (PowerShell 7+):

Set-ExecutionPolicy Bypass -Scope Process -Force; $wc=New-Object Net.WebClient; $wc.Encoding=[Text.Encoding]::UTF8; iex $wc.DownloadString('https://higress.ai/hiclaw/install.ps1')

After installation, access http://127.0.0.1:18088, log in via Element Web, and the Manager will guide you through creating your first Worker. Interaction is conversational: tell the Manager to create Workers or @mention Workers to assign tasks directly in Matrix rooms.

Core Components

ComponentRoleLanguage
Higress AI GatewayLLM proxy, MCP Server hosting, credential managementGo
TuwunelSelf-hosted Matrix Homeserver
Element WebBrowser IM clientWeb
MinIOCentralized object storage
OpenClawAgent runtime (Matrix plugin + Skills)Node.js 22
CoPawLightweight agent runtime (~150MB RAM)Python 3.11
hiclaw-controllerKubernetes custom controller

⚠️ Note: The repository entry URL was originally alibaba/hiclaw, now redirecting to agentscope-ai/HiClaw. The official website still references alibaba/hiclaw. The exact relationship between the two is not explicitly stated in the README.

Related Projects

View All arrow_forward

STAY UPDATED

Get the latest AI tools and trends delivered straight to your inbox. No spam, just intelligence.

rocket_launch