How to Setup Openclaw With LMStudio

Source: Dev.to

Introduction

OpenClaw has generated a lot of buzz, evolving from Clawdbot → Moltbot → OpenClaw. Most tutorials rely on external APIs (OpenAI, Anthropic, Google, etc.), which can become expensive. This guide shows how to run OpenClaw locally using LMStudio on a Linux‑based Lenovo ThinkPad.

Installing LMStudio

1. Install LMStudio on your Linux system.
   If you need help, a YouTube tutorial can guide you through the installation process.

Selecting a Model

Because of limited hardware resources, a quantized version of GLM‑4.7 Flash was chosen. After downloading the model, LMStudio’s chat interface responded to a simple “hello” in about 50 seconds, which is slow but acceptable for initial testing.

Installing OpenClaw

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

During the installation, the manual configuration wizard was used. Some required fields (skills, model provider, token, etc.) were missing, so the configuration file needed manual editing.

Editing openclaw.json

Open ~/.openclaw/openclaw.json (or the path shown by the installer) and add the following sections. Adjust paths and values as needed for your environment.

{
  "meta": {
    "lastTouchedVersion": "2026.1.29",
    "lastTouchedAt": "2026-01-31T02:01:52.403Z"
  },
  "wizard": {
    "lastRunAt": "2026-01-31T02:01:52.399Z",
    "lastRunVersion": "2026.1.29",
    "lastRunCommand": "onboard",
    "lastRunMode": "local"
  },
  "models": {
    "providers": {
      "lmstudio": {
        "baseUrl": "http://127.0.0.1:1234/v1",
        "apiKey": "lm-studio",
        "api": "openai-responses",
        "models": [
          {
            "id": "glm-4.7-flash",
            "name": "GLM-4.7 Flash",
            "reasoning": true,
            "input": ["text"],
            "cost": {
              "input": 0,
              "output": 0
            },
            "contextWindow": 20000,
            "maxTokens": 8192
          }
        ]
      }
    }
  },
  "agents": {
    "defaults": {
      "model": {
        "primary": "lmstudio/glm-4.7-flash"
      },
      "workspace": "/home/Ubuntu/.openclaw/workspace",
      "compaction": {
        "mode": "safeguard"
      },
      "maxConcurrent": 4,
      "subagents": {
        "maxConcurrent": 8
      }
    }
  },
  "messages": {
    "ackReactionScope": "group-mentions"
  },
  "commands": {
    "native": "auto",
    "nativeSkills": "auto"
  },
  "hooks": {
    "internal": {
      "enabled": true,
      "entries": {
        "session-memory": {
          "enabled": true
        }
      }
    }
  },
  "gateway": {
    "port": 18789,
    "bind": "loopback",
    "mode": "local",
    "auth": {
      "mode": "token",
      "token": "generate-your-token"
    },
    "tailscale": {
      "mode": "off",
      "resetOnExit": false
    }
  },
  "skills": {
    "install": {
      "nodeManager": "npm"
    }
  }
}

Generating a Token

Create a token for the gateway authentication:

openssl rand -hex 20

Replace "generate-your-token" in the gateway.auth.token field with the generated value.

Verifying the Installation

Run the setup verification command:

openclaw setup

Expected output:

Config OK: ~/.openclaw/openclaw.json
Workspace OK: ~/.openclaw/workspace
Sessions: OK: ~/.openclaw/agents/main/sessions

Starting the Gateway

Check the gateway status:

openclaw gateway status

You should see a line similar to:

Listening: 127.0.0.1:18789

This confirms that OpenClaw is listening locally on port 18789.

Next Steps

At this point OpenClaw is installed and reachable via the local gateway. Future work includes:

• Interacting with the bot through the OpenClaw CLI or a compatible client.
• Adding custom skills or agents.
• Monitoring performance and adjusting model parameters as needed.

Stress‑Testing AI Agents (Optional)

If you develop AI agents for internal or commercial use, you can perform a quick stress test with the Zeroshot tool:

zeroshot scan --target-url https://your-target-url --max-attacks 20

The tool can run up to 50 attacks (or more) using a library of 1,000+ attack vectors across various AI system categories. Visit the Zeroshot website for a free trial.