OpenClaw Manager Deployment Guide

Introduction

OpenClaw Manager is a cross-platform visual management tool that helps you quickly deploy and manage OpenClaw AI bot gateways.

Zero-config deployment — Single executable file, auto-installs all dependencies
Web management UI — Create, start/stop, delete instances and switch models via browser
Multi-channel support — Telegram / Feishu / Discord
Multi-model support — GPT-5 series, Claude 4.5/4.6 series, Gemini series with automatic failover
Background mode — Supports --daemon mode, unaffected by SSH disconnection
Multi-language UI — Chinese / English / Japanese

Prerequisites

Before you begin, please ensure:

APIMart API Key obtained Log in to APIMart Console to get your API key (starts with sk-)
Messaging platform Bot created Prepare your Bot credentials based on the platform you want to use (see channel configuration section below)

Tip: If you don’t have an APIMart account yet, please register at APIMart and obtain an API key first.

Step 1: Download

Download the executable for your platform from GitHub Releases:

Platform	Filename
Windows x64	`openclaw-manager-win-x64.exe`
macOS ARM (Apple Silicon)	`openclaw-manager-macos-arm64.zip`
macOS Intel	`openclaw-manager-macos-x64.zip`
Linux x64	`openclaw-manager-linux-x64`
Linux ARM64	`openclaw-manager-linux-arm64`

Step 2: Run

Windows
Linux
macOS

Windows Deployment

Prerequisites

Windows 10/11 (64-bit)
Administrator privileges (required for creating Windows scheduled tasks to manage Gateway)

Launch

Right-click openclaw-manager-win-x64.exe → Run as administrator

On first launch, the following will be done automatically:

Check and install Node.js v22 (auto-downloads MSI for silent installation if not found)
Install OpenClaw CLI (via npm install -g openclaw)
Auto-open browser to the management UI

Must run as administrator, otherwise Windows scheduled tasks cannot be created to manage Gateway processes.

Linux Deployment

Prerequisites

Ubuntu 20.04+ / Debian 11+ / CentOS 8+ (x64 or ARM64)
root privileges (recommended)

Launch

Download and grant execute permission:

chmod +x ./openclaw-manager-linux-x64

Foreground mode (for testing):

./openclaw-manager-linux-x64

Background mode (recommended for production):

./openclaw-manager-linux-x64 --daemon

Daemon Management

# Start in background
./openclaw-manager-linux-x64 --daemon
# Or shorthand
./openclaw-manager-linux-x64 -d

# Check status
./openclaw-manager-linux-x64 --status

# Stop
./openclaw-manager-linux-x64 --stop
# Or shorthand
./openclaw-manager-linux-x64 -s

Firewall Configuration

If remote access is needed, open the following ports:

# Manager Web UI
ufw allow 51942/tcp

# Gateway ports (one per instance, starting from 18789)
ufw allow 18789/tcp

Tip: On Linux, Manager automatically binds to 0.0.0.0, allowing remote access. Log file is at ~/openclaw-manager.log.

macOS Deployment

Prerequisites

macOS 12+ (Apple Silicon or Intel)

Launch

Extract the downloaded zip file to get OpenClaw Manager.app
Double-click to run OpenClaw Manager.app
First run may require security settings:

System Settings → Privacy & Security → Allow Anyway

Tip: On macOS, Manager binds to 127.0.0.1 (local access only). Daemon commands are the same as Linux.

Step 3: Access Management UI

After the program starts, access the Web management UI:

http://127.0.0.1:51942

Linux remote access: Replace 127.0.0.1 with your server IP, e.g. http://your-server-ip:51942

Step 4: Create Instance

In the Web management UI, click + New Instance to start:

4.1 Set Instance Name

Enter an instance name (letters, numbers, underscores, hyphens only, 1-64 characters).

4.2 Select AI Model

Choose the AI model to use and enter your APIMart API Key (starts with sk-). Supported Models:

Model ID	Name	Features
`gpt-5`	GPT-5	Latest and most powerful
`gpt-5.1`	GPT-5.1	Enhanced version
`gpt-5.2-codex`	GPT-5.2 Codex	Code specialized
`gpt-5.2-pro`	GPT-5.2 Pro	Professional version
`claude-opus-4-6-20260320`	Claude Opus 4.6	Strongest reasoning
`claude-sonnet-4-6-20260320`	Claude Sonnet 4.6	Balanced choice
`claude-opus-4-5-20251101`	Claude Opus 4.5	Advanced reasoning
`claude-sonnet-4-5-20250929`	Claude Sonnet 4.5	Excellent for code
`claude-haiku-4-5-20251001`	Claude Haiku 4.5	Fast response
`gemini-2.5-flash`	Gemini 2.5 Flash	Fast multimodal
`gemini-3-pro-preview`	Gemini 3 Pro Preview	High performance

Model Recommendations:

💰 Cost-effective: claude-haiku-4-5-20251001, gemini-2.5-flash
🚀 High-performance: gpt-5, claude-opus-4-6-20260320, gemini-3-pro-preview
⚡ Fast response: gemini-2.5-flash, claude-haiku-4-5-20251001

You can also select failover mode to add multiple backup models. When the primary model is unavailable, the system automatically switches to backup models.

4.3 Configure Messaging Channel

Select your messaging platform and follow the corresponding tutorial:

Create a Telegram bot via BotFather

Feishu

Create a Feishu enterprise bot

Discord

Create a Discord server bot

Step 5: Pairing Code Binding

After the instance is created and started, complete user binding via pairing code:

User sends any message to the bot on the messaging platform
Bot replies with an 8-digit pairing code (e.g. DFE62DTD)
Admin clicks the “Pairing Code” button for that instance in the Web UI
Enter the pairing code and click “Approve”
User can now use the bot normally

Tip: Pairing codes are stored in Gateway memory and expire after restart. If a code expires, simply have the user send a new message to get a fresh code.

Manage Instances

In the management UI, you can perform the following operations:

Action	Description
Start	Start the Gateway process, bot begins receiving messages
Stop	Stop the Gateway process
Switch Model	Switch AI model in real-time without restarting Gateway
Pairing Code	Approve new users’ pairing codes to authorize bot usage
Delete	Stop and delete instance with all data (irreversible)

Ports & Data

Ports

Purpose	Port	Bind Address
Manager Web UI	`51942`	Windows/macOS: `127.0.0.1`, Linux: `0.0.0.0`
Gateway Instance	Starting from `18789`	Each instance gets a unique port

Data Directories

Content	Path
Instance config	`~/.openclaw-<instance-name>/openclaw.json`
Gateway logs	`~/.openclaw-<instance-name>/logs/gateway.log`
Manager logs	`~/openclaw-manager.log` (daemon mode)
Manager PID	`~/.openclaw-manager.pid`

FAQ

Q1: Manager won’t start?

Solutions:

Platform	Check
Windows	Confirm running as administrator
Linux	Confirm port 51942 is not in use (`lsof -i :51942`)
macOS	Confirm security settings allowed or quarantine removed
All	Check logs at `~/openclaw-manager.log`

Q2: openclaw command not found?

Solutions:

The program automatically detects and installs openclaw CLI on startup
If version is too old (< 0.1.0), it will auto-update
Manual install: npm install -g openclaw@latest

Q3: Instance shows “stopped” after starting?

Solutions:

Gateway takes a few seconds to bind the port, status may briefly appear inconsistent
Refresh the page to check the latest status
Check Gateway logs: ~/.openclaw-<instance-name>/logs/gateway.log

Q4: Feishu bot not responding?

Solutions:

Confirm event subscription has added im.message.receive_v1
Confirm subscription method is “Long Connection”
Confirm a new version has been created and published (required after every permission/event change)

Q5: Pairing code approval failed?

Solutions:

Confirm Gateway is running
Pairing codes are stored in Gateway memory and expire after restart
Have the user send a new message to get a fresh pairing code

Q6: How to view API usage and costs?

📊 API call statistics
💰 Cost details
📈 Usage trend charts

Support & Help

If you encounter any issues:

Chat Applications

Development Tools

Application Platforms

​Introduction

​Prerequisites

​Step 1: Download

​Step 2: Run

​Windows Deployment

​Prerequisites

​Launch

​Linux Deployment

​Prerequisites

​Launch

​Daemon Management

​Firewall Configuration

​macOS Deployment

​Prerequisites

​Launch

​Step 3: Access Management UI

​Step 4: Create Instance

​4.1 Set Instance Name

​4.2 Select AI Model

​4.3 Configure Messaging Channel

Telegram

Feishu

Discord

​Step 5: Pairing Code Binding

​Manage Instances

​Ports & Data

​Ports

​Data Directories

​FAQ

​Q1: Manager won’t start?

​Q2: openclaw command not found?

​Q3: Instance shows “stopped” after starting?

​Q4: Feishu bot not responding?

​Q5: Pairing code approval failed?

​Q6: How to view API usage and costs?

​Support & Help

Get Started with APIMart

Introduction

Prerequisites

Step 1: Download

Step 2: Run

Windows Deployment

Prerequisites

Launch

Linux Deployment

Prerequisites

Launch

Daemon Management

Firewall Configuration

macOS Deployment

Prerequisites

Launch

Step 3: Access Management UI

Step 4: Create Instance

4.1 Set Instance Name

4.2 Select AI Model

4.3 Configure Messaging Channel

Step 5: Pairing Code Binding

Manage Instances

Ports & Data

Ports

Data Directories

FAQ

Q1: Manager won’t start?

Q2: openclaw command not found?

Q3: Instance shows “stopped” after starting?

Q4: Feishu bot not responding?

Q5: Pairing code approval failed?

Q6: How to view API usage and costs?

Support & Help