Run multiple OpenClaw agents from one self-hosted dashboard

OpenClaw is an open-source runtime that gives AI agents a browser, terminal, and file system to work with. It handles the workspace for each agent. But managing many workers takes visibility, coordination, and control.

That’s where Claworc comes in:
OpenClaw runs agents. Claworc helps you operate them.

One workspace for every agent

Each Instance gives an OpenClaw agent its own isolated workspace — with chat, a live browser, terminal, file manager, and persistent storage.

Webchat built in

Chat with the agent while watching it work

The webchat runs side by side with the embedded browser, so you can send instructions and see the agent act on them in real time.

Open the dashboard and export this week's report

→ navigating to /reports

claworc.app / instances / research-agent-04

Live browser

Watch or control the agent's browser as it works.

Terminal

Run command-line operations inside the workspace.

File manager

Browse, upload, and edit files.

Persistent storage

Keep files across restarts and redeployments.

Each instance runs in an isolated workspace with its own file system. Claworc monitors them continuously and restarts them automatically if they crash.

What You Can Do

Manage your entire fleet of AI agents from one place.

Create & Manage Instances

Spin up new agent workspaces, start/stop them, or remove them when done.

Documentation

research-agent-04 running

data-scraper-02 starting

email-agent-01 stopped

Chat with Agents

Send instructions and have a conversation with the AI agent in each instance.

Watch the Browser

See what the agent is doing in Chrome in real time, or take control yourself.

Manage Files

Browse and manage the files in each instance's workspace.

View Logs

Stream live logs to monitor what's happening inside an instance.

Skills Library

Package reusable prompts as skills and deploy them to instances in one click. Discover community skills on Clawhub.

Documentation

web-researcher form-filler ↗ clawhub report-gen pdf-parser ↗ email-sender

Configure Models & API Keys

Each agent gets a virtual key — real API credentials stay in the control plane and are never exposed inside containers.

Claworc keeps team access simple: admins manage all OpenClaw instances, while users only see the workspaces assigned to them.

Agents connect to LLM providers through Claworc using virtual keys, so real API credentials never leave the control plane. Every request is tracked by instance, provider, token usage, and cost.

Workspace data, handled safely

Share files between instances in real time, create snapshots on demand or on a schedule, and restore workspaces when needed.

Instance A

Instance B

Instance C

Shared Folder

Files sync across connected instances

Snapshot

Manual or scheduled backup

Download or restore

Use locally or restore to any running instance

Real-time sharing

Files appear across connected instances immediately.

Flexible mapping

Choose which instances share each folder.

Scheduled backups

Back up one, several, or all instances automatically.

One-click restore

Restore to any running instance.

Get Started

Claworc runs on Docker for local setups or Kubernetes for production. The control plane is a single 20 MB binary.

curl -fsSL https://raw.githubusercontent.com/gluk-w/claworc/main/install.sh | bash

Deploy with Helm For Kubernetes production deployments with full control over configuration.

Docker Compose For local development or single-server setups.

Documentation Architecture, API reference, data model, and more.

---

Claworc is fully open source, self-hosted, and free. GitHub · Discord · Twitter · Contributions welcome