Create a Hermes Agent in the AI Management Center, configure models, and integrate with DingTalk for IM-based command dispatching.
Prerequisites
Account and authentication
You have activated an Alibaba Cloud account and completed your profile information.
Create end users
Follow Create a convenience account to quickly create a convenience account, or see Set up user accounts to select an account type that fits your enterprise.
Obtain the logon credentials, including the organization ID, office network ID, username, and password.
Create a Hermes Agent
Go to the AI Management Center.
Click Create Cloud Hermes Agent.
Select a plan based on your business requirements:
Plan
Credit quota
Starter
200 credits/month
Standard
1,600 credits/month
Premium
4,000 credits/month
NotePlan prices vary by region. The actual cost is displayed on the cloud desktop purchase page. We recommend that you consider both network latency (select a nearby region) and price when choosing a region.
Credits measure the computing resources consumed when AI executes tasks. The actual consumption depends on task complexity and the model used. For estimated consumption per feature and model, see Credit consumption reference.
When your monthly credits are exhausted, the Hermes Agent cannot continue using the built-in plan models, even if the plan is still active. You can wait for the next monthly credit reset, subscribe to a Wuying shared credit package, or Configure custom models.
After the plan expires, the Hermes Agent cannot be used even if credits remain. You must renew the plan to continue using the service.
Select the Quantity and Duration, and specify whether to enable Auto-renewal.
Click Buy Now, confirm the order, and complete the payment.
Configure models for a Hermes Agent
Hermes Agent plans include preconfigured models ready to use after purchase. If your monthly plan credits are exhausted, follow these steps to add models from a shared credit package.
Create a model group
On the Hermes Agent page, click the Model Configuration tab.
Click Create Model Group, enter a model group name, and complete the creation.
Add models
Select the Model Group you created, and click Add Model.
NoteEach model group supports up to 50 models.
Click Add Model. In the panel, select a model integration method. The following three methods are supported:
Wuying Model: Offers a range of mainstream large language models at competitive prices. Supported models include Qwen Max, Qwen Plus, Qwen Coder, Qwen Flash, DeepSeek, Kimi, GLM, MiniMax, and more.
Wuying Package Model: Consumes credits included in your plan. No additional purchase required.
Wuying Credit Package Model: Requires a separate Wuying shared credit package subscription.
Model Square: Integrates with model service providers such as Alibaba Cloud Model Studio.
Custom Model Configuration: Adds your own model endpoints. Supports OpenAI API and Anthropic API compatible modes.
For detailed configuration steps for each provider, see Configure custom models.
(Optional) Select a model and click the Set as Default Model button on its card.
NoteIf no default model is set, the system uses the first model added as the default.
Assign a Hermes Agent to a user
On the Hermes Agent page, click More > View/Add Users in the Actions column for the target Hermes Agent.
In the View/Add Users panel, click Add User, search for the target user, and complete the assignment.
Each Hermes Agent supports only one user. To change the assigned user, remove the current user first and then add a new one.
Integrate with DingTalk and chat with the bot
Hermes Agent supports IM platform integration for dispatching commands through chat. The following example uses DingTalk. For other integration options, see More channel integrations.
Go to the AI Management Center.
Click the Channel Configuration tab.
In the Actions column of the target agent, click View/Configure Channel.
On the DingTalk card, click Configure Now.
In the dialog box, select a method to complete the bot integration:
Quick pairing
The cloud desktop image version must be Linux 3.0.10 or later. To update, restart the cloud desktop.
Click Send Now. The system sends an SMS or email containing an access link to the authorized user.
The authorized user opens the URL from the SMS or email to access the binding page and obtain a QR code (you can also copy the URL directly from the console).
Scan the QR code with DingTalk to bind the bot.
Chat with the bot in DingTalk to verify the connection.
If another bot is already bound, scanning the code will replace it.
Link an existing bot
Create a DingTalk application
Your DingTalk organization must have developer permissions. If your organization does not have this permission, see Obtain developer permissions. If you do not have a suitable organization, scan the following QR code with the DingTalk mobile app to quickly create one.
Log on to the DingTalk Open Platform developer console.
In the top navigation bar, click App Development. On the App Development page, click Create Now to create an OpenClaw bot application with one click.
On the Create OpenClaw page, enter basic information such as the bot name, description, and icon. You can also keep the default settings and click OK to complete the creation.
After creation, the page displays the Client ID and Client Secret for the application. Copy and save them immediately — you will need them when configuring the DingTalk channel in the AI Management Center.
NoteThe Client ID and Client Secret are core credentials for accessing application data. Keep them secure and do not share them.
Enter credentials to complete the linkage
Go back to the AI Management Center, enter the Client ID and Client Secret of the DingTalk application, and click OK.
Chat with the bot in DingTalk to verify the connection.
Chat with the DingTalk bot
Open DingTalk and enter the name of the DingTalk application you created in the search box at the top.
In the search results, click the bot under the Features section to open the chat.
Enter a question or command to start chatting with the DingTalk bot.
FAQ
I cannot find the Hermes Agent entry in the Alibaba Cloud Workspace client after purchase
Hermes Agent is not accessible from Alibaba Cloud Workspace clients. Connect it to an IM platform such as DingTalk to interact through chat. For integration methods, see More channel integrations.
An error indicating an invalid API Key or verification failure occurs when adding a model
Go to the API Key management page in the Alibaba Cloud Model Studio console and recopy the full API Key (starts with
sk-). Ensure no prefix or trailing characters are missing.Check the API Key status to confirm it has not been disabled or expired. If expired, create a new API Key.
Confirm that your Alibaba Cloud Model Studio account has sufficient balance. If overdue, top up your account first.
After channel configuration, the DingTalk bot does not respond or returns an authentication failure
On the DingTalk Open Platform, verify that the Client ID and Client Secret under Credentials and Basic Information match the values in your channel configuration.
Model invocation errors or no response when chatting with the Hermes Agent
On the model configuration tab, confirm that a model group has been created and contains at least one available model.
If using a manually added model, verify that the API Key is correct and the corresponding model service is running. You can test the API Key directly in the Alibaba Cloud Model Studio console.
On the Hermes Agent page, check the credit usage details to confirm whether your plan credits are sufficient. If credits are insufficient, upgrade your plan or purchase additional credits.
What to do next
More channel integrations
Configure custom models
Add your own models through Model Square or Custom Model Configuration.
Model Studio pay-as-you-go
In the AI Management Center, click Add Model.
Select Model Square > Alibaba Cloud Model Studio.
Based on the region of your API Key, select the corresponding option from the Base URL drop-down list:
China (Beijing):
https://dashscope.aliyuncs.com/compatible-mode/v1Singapore:
https://dashscope-intl.aliyuncs.com/compatible-mode/v1US (Virginia):
https://dashscope-us.aliyuncs.com/compatible-mode/v1
Enter the API Key.
Model Studio Token Plan
Go to the Token Plan Team Edition purchase page. Select the seat type and quantity, and complete the subscription.
API Key: On the Token Plan Team Edition page, copy your dedicated API Key.
In the AI Management Center, click Add Model.
Select Model Square > Model Studio Token Plan.
Select the compatibility mode and corresponding Base URL from the table below:
Compatibility mode
Base URL
OpenAI compatible
https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1Anthropic compatible
https://token-plan.ap-southeast-1.maas.aliyuncs.com/apps/anthropicEnter the API Key.
For models supported by Alibaba Cloud Model Studio Token Plan, see Overview.
Model Studio Coding Plan
Subscribe to Alibaba Cloud Model Studio Coding Plan and generate and save an API Key.
In the AI Management Center, click Add Model.
Select Custom Model Configuration.
Select the compatibility mode and corresponding Base URL from the table below:
Compatibility mode
Base URL
OpenAI compatible
https://coding-intl.dashscope.aliyuncs.com/v1Anthropic compatible
https://coding-intl.dashscope.aliyuncs.com/apps/anthropicEnter the API Key.
For models supported by Alibaba Cloud Model Studio Coding Plan, see Overview.
Other providers
The Model Square offers providers such as SiliconFlow, Moonshot AI, MiniMax, and Zhipu AI, which you can select and configure directly. If the provider you need is not listed in Model Square, you can use Custom Model Configuration to integrate it. Follow these steps to configure a custom model:
Obtain the Base URL and API Key from the corresponding provider.
In the AI Management Center, click Add Model.
Select Custom Model Configuration.
Enter the Model Provider Name and Model Provider ID.
Select the compatibility mode (OpenAI or Anthropic), and enter the Base URL and API Key.