OpenClaw with APIMaster.ai

OpenClaw is a local AI agent platform with a web chat UI. Connect APIMaster.ai (OpenAI-compatible) through openclaw onboard → Custom Provider.

Get API Key first. Use placeholder your_apimaster_key below.

---

Prerequisites

• Windows, macOS, or Linux
• Node.js 22.19+ (Node 24 recommended; the installer upgrades if needed)
• APIMaster API Key ready

---

Step 1 — Install

After install you usually get Starting setup… (onboard). To install only and configure later, see Install without onboard below.

Windows (PowerShell)

powershell -c "irm https://openclaw.ai/install.ps1 | iex"

macOS (Terminal)

1. Open Terminal.app (or iTerm).
2. Run the official installer (OpenClaw docs):

curl -fsSL https://openclaw.ai/install.sh | bash

On macOS the script will:

• Detect Apple Silicon vs Intel
• Install Homebrew if missing (requires an Administrator account)
• Install Node 24 (or another build ≥ 22.19) via Homebrew when needed
• Install the openclaw CLI and usually start onboard

Check Node first (optional):

node --version   # need v22.19+, v24.x recommended

Install without onboard:

curl -fsSL https://openclaw.ai/install.sh | bash -s -- --no-onboard

Then run openclaw onboard when ready.

macOS config paths:

Item	Path
Config	~/.openclaw/openclaw.json
Workspace	~/.openclaw/workspace
Sessions	~/.openclaw/agents/main/sessions

Screenshots below are from Windows, but the CLI onboard flow is the same on macOS (Custom Provider, Base URL, Skip channel, etc.).

Linux / WSL2

Same as macOS:

curl -fsSL https://openclaw.ai/install.sh | bash

Optional background Gateway:

openclaw onboard --install-daemon

(macOS: LaunchAgent · Linux: systemd user service)

---

Step 2 — openclaw onboard

Run anytime to (re)configure:

openclaw onboard

On macOS, use the same Terminal session. For a persistent Gateway:

openclaw onboard --install-daemon

1. Security notice → Yes
2. Model/auth provider → Custom Provider (OpenAI or Anthropic compatible endpoint)

3. Fill in:

Field	Value
API Base URL	https://apimaster.ai/v1
API key	Paste APIMaster key
Endpoint compatibility	OpenAI-compatible
Model ID	e.g. gpt-5.5

Model ID: any marketplace chat model id (gpt-5.4, claude-sonnet-4-6, …). Skip image models like gpt-image-2.

4. Optional steps: Skip channel, web search, skills, hooks.

---

Step 3 — Web chat

If the browser does not open:

openclaw dashboard

If the page fails to load, run openclaw dashboard again or re-run openclaw onboard.

---

Step 4 — Test

Pick the model under the input (e.g. gpt-5.5 (Custom Provider)), send hello, confirm a reply.

---

Troubleshooting

Issue	Action
Bad config	openclaw onboard
Page won't open	openclaw dashboard
401 / model error	Check URL, key, model id; connectivity test

macOS

Issue	Action
openclaw: command not found	Add to ~/.zshrc: export PATH="$(npm prefix -g)/bin:$PATH", reopen Terminal
Node version still wrong with nvm	nvm install 24 && nvm use 24 && nvm alias default 24, rerun installer
Homebrew install fails	Use an Administrator account, or install Homebrew first
Gateway not running	openclaw gateway status; try openclaw onboard --install-daemon

---

Checklist

• Custom Provider + https://apimaster.ai/v1 + key
• Chat model id (not image)
• openclaw dashboard works
• Test message succeeds

---

See also

• Agents overview
• OpenClaw docs · Install (macOS/Linux)