Configuring the OpenClaw Application

The OpenClaw application image provided by FlexusL can be integrated with multiple mainstream AI models. You can easily build a personal AI assistant by connecting it to Telegram, Discord, and WhatsApp. For more best practices, see Summary of OpenClaw Best Practices.

If you need to configure and manage OpenClaw, you can perform operations described in this section on the FlexusL console.

Prerequisites

You have purchased a FlexusL instance with the OpenClaw application image installed.

Application Management

Log in to the FlexusL console.
Click Configure Now on the target FlexusL instance. The application details page is displayed.
If Configure Now is not displayed, click the resource card to go to the resource details page and choose Application Details from the left navigation pane.

Some functions of the OpenClaw application image (such as querying gateway status and updating models and application configurations) depend on Cloud Operations Center (COC). You need to enable and authorize COC before using these functions. If the content in the following figure is displayed, enable and authorize COC as prompted.

In the Application Management card of the Basic Configurations tab, view the application details and manage the application as needed.

If the page has not been refreshed for a long time or you need to check whether the gateway status has been changed, click to view the latest version and status.

**Table 1** Application details
Application Details	Example	Description
Version	2026.3.2	OpenClaw version.
Status	Running	OpenClaw gateway status. Running: The OpenClaw service is working normally. Unknown/Unhealthy: The initial status of the OpenClaw gateway is Loading. If the status changes to Unknown or Unhealthy after several minutes, click Restart. The restart takes some time. After the gateway is restarted successfully, the status changes to Running. NOTE: If the gateway status is still Unknown or Unhealthy after you click Restart, submit a service ticket.

Application management operations

OpenClaw Web UI access

If you want to use the OpenClaw Web UI to manage OpenClaw, click Enable. Confirm security risks and click OK.

Enabling public IP address access to the OpenClaw Web UI may expose your server to network attacks. After enabling this function, configure the security group rules to reduce risks.
Click OpenClaw Web UI access. In the displayed dialog box, a message is displayed, requiring you to keep the information secure to prevent privacy leakage or asset loss.
- Web UI Access URL: The URL for accessing the OpenClaw Web UI.
- Port: The gateway port of the OpenClaw Web UI.
  
  To prevent the server from being maliciously scanned, the system will change the default port 18789 to another unused port and enable access to this port. In this case, the port's source IP address in the security group inbound rule is 0.0.0.0/0, indicating that all external IP addresses are allowed to access your server. This may cause severe consequences such as network intrusion, service interruption, data breach, or data ransomware. You should only configure trusted IP addresses for the source in security group rules.
  
  In the upper right corner of the page, click Configure Security Group Rule. In the inbound rules, find the target port and change the source IP address of the port to a trusted IP address.
- Token: The token of the OpenClaw Web UI. To prevent malicious access, you need to add the token when accessing the UI.
Click Access Now or the Web UI access URL to open the OpenClaw Web UI.
The UI is provided by OpenClaw. For details, see the OpenClaw official documentation.
After you finish the operations, disable OpenClaw Web UI access. Once disabled, the OpenClaw Web UI will be inaccessible and the security group rules that allow access to the port will be automatically deleted. This can reduce data breach and other security risks.

Version Upgrade

If the OpenClaw gateway status is Running and a new version is available, click Version upgrade.
If the OpenClaw gateway status is Running and Version upgrade is grayed out, the current version is already the latest.

For more details, see Upgrading the OpenClaw Version.

Auto-repair

If OpenClaw cannot be started due to incorrect API keys, plug-ins, or channel configurations in the OpenClaw configuration files, click Auto-repair. The system automatically runs the OpenClaw repair command (openclaw doctor --repair --force) to detect and repair the fault.
After auto-repair is complete, it only indicates that the repair command has been executed, but the OpenClaw gateway status may not be Running. If the OpenClaw gateway status is still Unknown or Abnormal, or the configured bot is unavailable, remotely log in to the FlexusL instance and run the commands below to view OpenClaw logs. If the fault persists, submit a service ticket.
```
su - openclaw
cd /tmp/openclaw
ll
```
Find the target log file and run the vi file-name command to view the log content.

Backup and Rollback

FlexusL supports manual, scheduled, and automatic backup of the openclaw.json configuration file. If a file modification causes an exception, you can quickly roll back to a specified historical version. For more information, see OpenClaw Backup and Rollback.

Restart

If the OpenClaw gateway status is Unknown or Abnormal and is not in the auto-repair or version upgrade state, you can click Restart to reload the OpenClaw gateway.

Stop/Enable

If the OpenClaw gateway status is Running and is not in the auto-repair or version upgrade state, you can click Stop to stop the running OpenClaw gateway.
After the gateway is stopped, OpenClaw will be inaccessible, the status becomes Abnormal, and the configured bot cannot be used.
After the gateway is stopped, you can click Enable to start OpenClaw again. When the OpenClaw gateway status is Running, OpenClaw becomes accessible and the configured bot can be used.

Model Configuration

Log in to the FlexusL console. Click Configure Now on the target FlexusL instance card and go to the application details page.

Configure required parameters in the Models card.

**Table 2** Model configuration parameters
Parameter	Example	Description
Model Provider	Huawei Cloud MaaS	Huawei Cloud MaaS Model APIs Huawei Cloud MaaS models are billed based on the actual usage. For details, see Billing Details. Custom If the AI models provided by Huawei Cloud MaaS do not meet your requirements, you can connect to custom models. For details, see Connecting OpenClaw with Custom AI Models.
API Key	HpkNxxxxxxxxxxxxxxxxxxxxxxxrHZw	You can use an existing API key created for the CN-Hong Kong region or create an API key for the CN-Hong Kong region as instructed. CAUTION: The API key will be displayed only once upon creation. Please copy and store it securely. If you lose it, you will need to create a new API key.

Click Check Connectivity. The system will check whether the models are connected and available based on the entered information. This function is available only prior to adding the first model.
Click Add Model. All available models will be displayed. You can select and add them in batches. If you are adding MaaS models for the first time, read and agree to the MaaS service statement.
Click OK. You can view the added models.
- When adding models for the first time, if you select Huawei Cloud MaaS, DeepSeek-V3.2 will be automatically set as the default model. You can change the default model if needed.
- If you have already added models and set a default model, any newly added model will not be used until you manually set it as the default.

You can perform the following operations on added models.

**Table 3** Model operations
Operation	Description
Modifying the API key	Click Modify next to API key. Then, the system will clear the configured API key. You can enter an API key already created for the CN-Hong Kong region or create an API key for the CN-Hong Kong region as instructed. After the configuration is complete, click Add Model. All available models that have not been added to Huawei Cloud MaaS are displayed. Click OK to update the API key and add models in batches.
Viewing call details	Click Call Details on the right of the target model. Huawei Cloud MaaS models: Go to the Huawei Cloud MaaS console to view the model call details, including the total number of calls, failed calls, tokens, input tokens, output tokens, and end-to-end latency. For more information, see Viewing the Call Data and Monitoring Metrics of Real-Time Inference in MaaS. Custom models: Obtain the call details from the model provider.
Modifying the model alias	Hover over on the right of the target model and click Modify Alias. In the displayed dialog box, specify a new alias that is easy to identify.
Changing the default model	Click Set as Default on the right of the target model to set it as the new default model. The new default model will not be used unless you start a new session. You can enter /new model-alias (start a new session) or /model model-alias (retain the context) in the chat box. CAUTION: When you enter /new in the chat box to start a new session, the bot will lose its memory. Exercise caution when performing this operation.
Deleting a model	Click Delete on the right of the target model. If the model to be deleted is the default model, you can choose to retain or forcibly delete it. If the default model is forcibly deleted, the system will randomly set another existing model as the new default model. You can change the default model if needed. CAUTION: After a model is deleted, the bot cannot use it anymore. To reuse the same model, add it again.

Channel Configuration

Log in to the FlexusL console. Click Configure Now on the target FlexusL instance card and go to the application details page.

In the Channels card, configure required parameters and click Add Bot.

**Table 4** Bot configuration parameters
Parameter	Example	Description
Channel	Telegram	Select the channels to be connected. Discord Telegram
Bot details	Bxxxxxxtoken	Bot credentials: See the corresponding best practices to obtain the required credentials. Step 2: Create a Discord Bot Step 2: Create a Telegram Bot NOTE: The WhatsApp channel cannot be configured on the current page. To configure it, go to Step 3: Integrate WhatsApp into OpenClaw.

After the bot is configured, view it in the connected channels list.
- You can add more bots as needed. A maximum of 10 bots are allowed for all channels.

Manage the added bots as needed by referring to the following table.

**Table 5** Bot management operations
Operation	Description	Procedure
Creating a team	You can create a bot team. By configuring the bot persona and instructions, you can enable multiple bots to collaborate with each other to complete tasks. For configuration examples, see OpenClaw Bot Team Collaboration Example. Currently, only one team can be created. The master bot in a team cannot be removed or deleted after being set.	On the Teams tab, click Create Team. Select the master bot and assistant bot, confirm the selection, and click OK.
Adding a member	You can select bots that are not yet in a team to join the team. Once added, they will become assistant bots. NOTE: WeCom bots do not support teams.	On the Teams tab, click Add Member. Bots that are not yet in a team are displayed in the list. Select one or more bots and click OK.
Removing a member from a team	Only assistant bots can be removed from a team.	On the Teams tab, choose > Remove on the right of the assistant bot to be removed.
Disbanding a team	After a team is disbanded, all members are removed from the team. All collaboration permissions are immediately terminated and cannot be restored.	On the Teams tab, click Disband on the right of the team. Confirm the operation and click OK.
Deleting a bot	After deletion, the bot will be no longer available. Exercise caution when performing this operation. If the bot to be deleted is the master bot of the team, the team will be automatically disbanded after the bot is deleted. If the bot to be deleted is the assistant bot of the team, it will be automatically removed from the team after being deleted.	On the Channels or Teams tab, choose > Delete on the right of the target bot.
Modifying an alias	You can customize the bot name for easy identification.	On the Channels or Teams tab, choose > Modify Alias on the right of the target bot.
Configuring bot basic information	You can modify the basic information about a bot.	On the Channels or Teams tab, click Set on the right of the target bot to modify the following basic information: Bot alias: Customize a name for easy identification. Bot credentials: including the App ID and App Secret. Model: You can specify a model that has been added. If no model is specified, the bot uses the default model. After specifying a model, you can view it in the model list and bot list. NOTE: After modifying the information, click Confirm for the modification to take effect.
Configuring bot basic information	In OpenClaw, the role of a bot is defined by a set of core configuration files, which determine its identity, personality, working style, and memory. Persona (SOUL.md) is the soul file that defines the identity, values, tone, and behavioral boundaries. Instructions (AGENTS.md) refer to the instruction manual that defines the workflow, operational guidelines, and permissions. These files are stored in the workspace directory of OpenClaw and are automatically loaded and applied during startups and sessions. FlexusL provides a bot template library, covering scenarios such as product R&D, sales and marketing, efficiency improvement, and finance. Multiple preconfigured bot assistant templates are available. You can customize your own bot roles by referring to the official OpenClaw templates and the bot templates provided by FlexusL. NOTE: FlexusL-provided bot templates are for reference only. You can manually modify the templates or use AI models to adjust them based on your specific service requirements.	On the Channels or Teams tab, click Set on the right of the target bot and customize the bot's persona, instructions, and private skills. Alternatively, click Templates and search for a suitable template from the template library provided by FlexusL. In the bot template library, click any template to view its persona, instructions, and skills. If the template meets your requirements, click Apply, select the target bot, and click OK. NOTE: After the bot template is applied, enter /new in the chat box to start a new session for the template to take effect. You need to configure multi-bot collaboration for the master bot in its instructions (AGENTS.md) to enable team collaboration. Therefore, the templates in the library cannot be applied to the master bot. You can apply templates to the master bot on the Channels or Teams tab and customize the settings after the application. For configuration examples, see OpenClaw Bot Team Collaboration Example. Persona and instruction settings support Markdown syntax. As you edit content in the left pane, a real-time preview displays the formatted output in the right pane. For example, to create a heading, add a number sign (#) in front of a word or phrase. The number of number signs indicates the heading level. Each number sign must be followed by a space. Skills installed on the bot configuration page are private to the current bot. This page does not display preconfigured shared skills or shared skills that have been installed in the skill list. To delete a private skill of the bot, click Confirm to make the deletion take effect.

Skill Installation

Log in to the FlexusL console. Click Configure Now on the target FlexusL instance card and go to the application details page.
In the Skills card, view the installed skills in the installed skills list, including the skills preconfigured in the OpenClaw application image.
Install official skills online. For details about the skills, see ClawHub Skills. Click the target skill to go to the skill details page and obtain the skill name from the CLI commands. For example, if you want to install Skill Vetter, obtain the skill name skill-vetter from the CLI commands. If you use Skill Vetter as the skill name, the installation will fail.

Some skills (such as Playwright MCP) require additional components to function. However, these dependencies are not automatically installed during skill installation. You need to install them manually. For details, see What Should I Do If a Skill Cannot Be Used Normally After Being Installed Online?
Enter the name of the skill to be installed in the text box and click Install Skills.
Confirm the skill name and click OK.
After the installation is successful, view the installed skills in the installed skills list.
Installed shared skills can be deleted. After a skill is deleted, all bots cannot use the skill.

If the system displays a message indicating that the ClawHub community is busy and the installation fails, see What Should I Do If Skills Failed to Be Installed?