AI Agent Workflows includes a DeepSeek chat action and registers DeepSeek through an OpenAI-compatible provider implementation. The available model list is retrieved from DeepSeek after the API key has been accepted.
Screenshot to add: DeepSeek chat action configuration with API key and model selection
Configure DeepSeek
- Open a workflow and add the DeepSeek Chat action, or add an AI Agent and connect a DeepSeek chat model.
- Select the node to open its configuration.
- Get an API key from the DeepSeek API keys page, then enter it in the provider API key field shown by the node.
- Wait for the key validation and model request to finish.
- Select a text-generation model from the returned list and configure the prompt and mapped inputs.
- Save the workflow, run a test, and review the node result.
Understand model availability
The plugin uses DeepSeek’s provider API to retrieve text-generation model metadata. It does not hard-code a permanent list of model names in this article. If the list is empty, check the key, provider access, and any error displayed in the editor.
Test before using the result
- Use representative input values and confirm the prompt receives the expected data.
- Review the output before mapping it into a later action.
- Do not assume that a valid key guarantees that every model request or complete workflow will succeed.