OpenClaw Setup (II): Install OpenClaw and Connect ECNU AI

Install OpenClaw locally, connect it to the ECNU AI platform, and verify that the environment works end to end.

Haoming Wang · Published March 15, 2026

Overview

In the first post, we collected the API Key, Base URL, and model name. The next step is to install OpenClaw locally and map those values into the initialization flow correctly.

OpenClaw × ECNU AI Setup Series
① Retrieve ECNU AI API credentials
② Install OpenClaw and connect it to the model (this post)
③ Connect a Feishu bot

View the series overview

If your goal is to reach a minimal working local environment as quickly as possible, this post covers exactly that path.

Goal of this post:
• Install OpenClaw locally
• Connect it to the ECNU AI platform through a custom provider
• Confirm that the model can respond correctly in your local environment

Run the installer

Step 1: Execute the installation command

Run the appropriate installation command for your operating system.

macOS / Linux

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

Windows PowerShell

iwr -useb https://openclaw.ai/install.ps1 | iex

Windows CMD

curl -fsSL https://openclaw.ai/install.cmd -o install.cmd && install.cmd && del install.cmd

Figure 1. Run the installation command in the terminal and wait for the download and initialization to complete.

Step 2: Confirm the installation prompt

Once the script finishes downloading the required components, OpenClaw opens its onboarding screen. Read the prompt and confirm that you want to continue.

Welcome screen — Figure 2. Review the onboarding prompt and confirm the installation.

Step 3-4: Choose QuickStart and switch to Custom Provider

No additional values need to be entered here. Simply follow the annotations in Figure 3 to enter the custom provider flow.

Connect the ECNU AI platform

Step 5-9: Complete model configuration within a single screen

From this point on, Figure 4 already contains steps 5 through 9 within a single interface. Follow the order shown in the screenshot:

Enter the Base URL recorded in the first post
Choose Paste API key now
Paste the saved API Key
Select Unknown (detect automatically)
Enter the model name you want to use, such as ecnu-reasoner or ecnu-max

Complete model configuration within a single screen — Figure 4. Steps 5-9 are grouped within one interface: fill in the Base URL, API key, and model name in order.

Finish initialization

Step 10-15: Complete the remaining options and enter the Web UI

After the model connection is in place, the rest of the onboarding flow mainly consists of optional settings. There is no need to over-optimize these choices during the first pass; follow the key selections marked in each screenshot.

If you are unsure about chat channels, search providers, or extra skills, it is fine to skip them for now
If OpenClaw asks for additional API keys you do not need, choose No
If you want a visual interface, select Open the Web UI in the final step

This part maps Figure 5 through Figure 10 to steps 10 through 15.

Choose a chat channel — Figure 5. Step 10: choose a chat channel. If you only need a local setup for now, select Skip for now.

Configure search capability — Figure 6. Step 11: the search provider can be skipped during the initial setup and added later if needed.

Choose skills — Figure 7. Step 12: select skills if you want them, or skip them to keep the setup minimal.

Handle extra key prompts — Figure 8. Step 13: if extra skill-specific API key prompts appear and you do not need them, choose No.

Configure hooks — Figure 9. Step 14: enable hooks only if they match your workflow; leaving the default is fine for a first setup.

Enter the Web UI — Figure 10. Step 15: choose Open the Web UI to enter the graphical interface.

Verify the result

After finishing the setup, ask a simple question in the interface. If the model responds correctly, the local installation and API connection are working as expected.

Verify the Web UI — Figure 11. Once the Web UI opens, ask a simple question to confirm that the model connection is working.

If something goes wrong: Check the three core values first: Base URL, API Key, and the model name. If those are correct but the onboarding flow became inconsistent, reinstalling is usually faster than unwinding the configuration manually.

At this point, you already have a usable local OpenClaw environment. The next step is to extend the same setup into Feishu.

The next post covers how to connect OpenClaw to a Feishu bot so that the same model can be used in mobile and group-chat scenarios.

OpenClaw Setup (III): Connect a Feishu Bot