HiClaw

What is HiClaw?

HiClaw is an open-source multi-agent operating system. It enables multiple AI agents to collaborate within Matrix chat rooms, with all communication fully visible to humans. The platform uses a Manager-Worker architecture to coordinate agents for completing complex tasks. It is designed to fast-track the creation of digital workforces for both individual entrepreneurs and enterprises.

Application scenarios

* Enterprise digital workforce: Building and managing a team of AI agents to handle complex business processes.
* OPOC (One Person One Company) acceleration: Enabling a solo entrepreneur to scale their operations by deploying a coordinated set of AI workers.
* Complex task automation: Decomposing and executing intricate projects that require parallel, specialized work.
* Secure AI development: Running multi-agent systems where security and credential protection are paramount.
* Auditable AI collaboration: Maintaining full transparency and logs of all interactions between AI agents for review and compliance.

Main features

* Multi-Agent Collaboration: A Manager Agent intelligently decomposes tasks and coordinates multiple Worker Agents for parallel execution.
* Human Supervision: Humans can enter any Matrix room at any time to observe conversations, intervene, or correct agent behavior in real-time.
* Enterprise Security: Worker agents never hold real API keys or GitHub PATs, using only a consumer token, preventing credential leaks even if compromised.
* Matrix Protocol Foundation: All agent communications occur on the open Matrix IM protocol, providing transparency, auditability, and support for distributed deployment.
* Manager-Worker Architecture: A clear two-tier system with distinct responsibilities for orchestration (Manager) and task execution (Worker), which is easy to extend with custom agents.
* Fully Open Source: The platform is community-driven, avoids vendor lock-in, and supports local private deployment for full data sovereignty.
* Higress AI Gateway: A central gateway provides secure external API access to LLM providers like Qwen, Claude, and OpenAI, as well as other external APIs and skills.
* Credentials Pass-through: A security layer ensures API keys are proxied securely and are never held directly by the agents themselves.

Target users

HiClaw primarily targets developers, engineers, and entrepreneurs looking to build and manage automated AI teams. It is suited for solo founders (OPOC) aiming to scale their capabilities and enterprises seeking to deploy secure, auditable digital workforces. The open-source nature also appeals to technical users who require full control and local private deployment.

How to use HiClaw?

The primary method is a local installation. For macOS, Linux, or Windows (with PowerShell 7+), you can run a one-command install script from your terminal. The script interactively asks for your LLM API key and sets up the system using Docker as the only prerequisite. Once installed, you access the Manager interface via a local web address, log in, and can begin instructing the Manager Agent to create and coordinate Worker Agents for tasks.

Effect Review

HiClaw presents a technically robust framework for multi-agent AI collaboration, with a strong emphasis on security, transparency, and user control. Its architecture, built on the open Matrix protocol and featuring a secure credential proxy, directly addresses critical concerns for deploying AI agents in sensitive or production environments. The ability for real-time human oversight and intervention provides a necessary safety layer, making it a practical solution for users who need more than just automated outputs but require managed, observable workflows. For developers and organizations comfortable with self-hosting, HiClaw offers a powerful and extensible foundation for building sophisticated AI-driven operations.