How to Install and Use OpenClaw AI Agent on Your Own Server
This guide explains what OpenClaw is, its capabilities and risks, the project’s rapid growth, and provides step‑by‑step instructions for installing the platform, configuring large language models, handling common errors, and connecting it to chat applications like DingTalk, Feishu, and others.
What OpenClaw Can Do
OpenClaw is an open‑source agent platform that runs locally and can be accessed from the chat apps you already use—WhatsApp, Telegram, Discord, Slack, Teams, etc. It acts as an AI assistant that follows you wherever you are.
Unlike the earlier Claw Code which only accepted a working directory and a bug‑fix command, OpenClaw provides multi‑agent capabilities, allowing you to delegate tasks such as file organization, email sending, or ordering food through chat tools like DingTalk, Feishu, or WeCom, effectively granting the AI full access to your machine.
Because of its extensive permissions, some institutions prohibit its installation on personal work computers; it is recommended to run it in a clean environment or use one of the vendor‑provided OpenClaw packages.
Background and Project Growth
OpenClaw has undergone three name changes—Clawdbot, Moltbot, and finally OpenClaw—to avoid confusion with Claude Code and retain the lobster branding. From its first commit on 2025‑11‑24 to 2026‑01‑31, the repository accumulated 8,297 commits (≈127 per day). The most active day recorded 349 commits, averaging one commit every 11 minutes, with 28.6% of activity occurring between 00:00–04:00, demonstrating a 7×24 development rhythm and earning over 80,000 GitHub stars within 24 hours.
OpenClaw Installation
Major cloud providers (Tencent, ByteDance, Alibaba) offer OpenClaw packages. After purchasing a lightweight server (e.g., Alibaba 2C2G for 79 RMB/year), you must choose a large language model. In China the main options are Alibaba’s Tongyi Qianwen Max (token‑based billing) and Zhipu GLM‑4.7 (monthly subscription).
If you use Tongyi Qianwen, follow the official tutorial to configure the app. For GLM‑4.7, obtain an API key from the Zhipu AI platform.
When the server image already includes OpenClaw 2026.2.3 or later, you can skip the token‑generation step. Otherwise, after logging into the server you run: openclaw onboard Key onboarding steps include: Onboarding mode:
QuickStart Model/auth provider:
Z.AI (GLM 4.7) Z.AI (GLM 4.7) auth method: enter your Zhipu AI API key Configure skills now?: select a couple of skills (can add later)
Skip the remaining API_KEY prompts Gateway service already installed: choose Restart After onboarding, configure the web section with your Brave search API key (free tier, requires a credit‑card for registration but no charge): openclaw configure --section web Enter the Brave API key when prompted, then verify the configuration on the Alibaba Cloud Gateway Dashboard.
Common Issues and Troubleshooting
⚠️ Note : If the server image is older, the token file path will be /root/.moltbot/moltbot.json , causing a “No such file or directory” error. Resetting the system resolves this.
If you encounter Failed to discover Alibaba Cloud models: 401 Unauthorized, you can ask OpenClaw itself for a solution; it will provide the necessary steps.
💡 Tip : For overseas deployments you must manually enable OpenClaw’s network access using a free Brave search API key.
Su San Talks Tech
Su San, former staff at several leading tech companies, is a top creator on Juejin and a premium creator on CSDN, and runs the free coding practice site www.susan.net.cn.
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.
