Jump to contentJump to page navigation: previous page [access key p]/next page [access key n]
documentation.suse.com / Configuring Open WebUI for AI Interaction

Configuring Open WebUI for AI Interaction

Publication Date: 19 Dec 2024
WHAT?

Open WebUI is a Web-based application that allows you to interact with AI models.

WHY?

To configure Open WebUI according to both user and administrator requirements and preferences.

EFFORT

To understand how to configure selected parts of Open WebUI takes less than 1 hour.

REQUIREMENTS

The procedures in this article assume that the following requirements are already fulfilled:

  • You have successfully deployed the SUSE AI.

  • You can access the Open WebUI interface in your Web browser.

  • Your Open WebUI user account has been created, and you have received the login credentials.

  • You have navigated your Web browser to the URL of your instance of Open WebUI and logged in with your credentials.

  • Selected tasks require the Open WebUI administrator privileges.

1 Introduction

 

This document provides an overview of Open WebUI and its user interface.

1.1 What is Open WebUI?

 

Open WebUI is a Web-based interface that allows you to interact with AI models, such as large language models (LLMs). It simplifies working with AI by providing a graphical user interface (GUI) instead of relying on command-line tools. The goal of Open WebUI is to make AI accessible to a broader audience, including those without technical expertise in AI development.

1.2 What are the key benefits of using Open WebUI?

 
Interaction with AI models

You can load a preferred AI model and interact with it using prompts in your natural language.

Custom data handling for RAG workflows

Open WebUI allows you to upload documents or specify URLs of pages that the AI can parse and reference during interaction, which is particularly useful for RAG workflows.

User-friendly interface

By providing a graphical user interface, Open WebUI makes AI tools more accessible to non-technical users. It helps them use the power of advanced AI models with no need for technical knowledge.

Data privacy

You can deploy Open WebUI locally or in a private environment. Such deployment is required for users concerned about privacy and data security.

1.3 Open WebUI user interface

 
Open WebUI user interface
Figure 1: Open WebUI user interface
 

The Open WebUI interface consists of three parts:

Top panel

Contains controls to manage tasks related to the current chat, such as:

  • Maximizing the chat window

  • Selecting an AI model for the current chat

  • Viewing and managing advanced settings for the current chat

  • Changing settings related to the current user

  • Viewing archived chats of the current user

  • Accessing the Admin Panel

Left sidebar

Contains links to perform the following tasks:

  • Start a new chat

  • Access the administrator's Workspace

  • Search the content of the current chat

  • Browse through the history of your chats

Main chat window

Displays your recent queries and corresponding answers provided by the AI. By using the prompt field at the bottom, you can accomplish the following tasks related to the current chat:

  • Enter the query that you want the AI to answer

  • Upload custom documents that you want the AI to process and use as a source of information when answering

  • Select a document, collection of documents, or URL to extend the AI model's knowledge base

  • Select a preconfigured AI model for the current chat

  • Interrupt the chatbot from answering if you are already satisfied with the provided information

1.4 Workspace

 
Important
Important

You must have Open WebUI administrator privileges to access configuration screens or settings mentioned in this section.

The Workspace area of the Open WebUI is a place where you can manage important parts of the AI stack. To access it, click Workspace at the top of the left panel. Workspace consists of the following parts:

Models

Manage existing AI modules or add your custom AI module based on an existing one. Refer to https://documentation.suse.com/suse-ai/1.0/html/openwebui-configuring/index.html#openwebui-managing-models for more details and related procedures.

Knowledge

Manage information sources—such as custom documents—that the users can use to enhance the current AI model's knowledge base. This helps making the model's responses more contextual. Refer to https://documentation.suse.com/suse-ai/1.0/html/openwebui-configuring/index.html#openwebui-managing-knowledgebase for more details and related procedures.

Prompts

Define the system prompts that guide the AI model to respond accurately in the style and format that you prefer. Refer to https://documentation.suse.com/suse-ai/1.0/html/openwebui-configuring/index.html#openwebui-managing-prompts for more details and related procedures.

Tools

Tools include a collection of Python snippets that enhance interaction and expand functionality for working with AI models. Tools can, for example, integrate additional data sources into the model knowledge base, such as databases.

Functions

Functions include a collection of Python snippets that can perform mostly internal data processing to automate workflows or manage text processing.

1.5 Admin Panel

 
Important
Important

You must have Open WebUI administrator privileges to access configuration screens or settings mentioned in this section.

The Admin Panel is a centralized control interface for administrators to manage system settings, user accounts and overall application functionality.

To access the Admin Panel, click your avatar button at the top right of the Open WebUI screen and select Admin Panel.

Admin Panel
Figure 2: Admin Panel
 

1.6 Known issues

 

This section lists the known issues with the Open WebUI and their solutions.

1.6.1 User password reset

 

An Open WebUI user can only reset their own password if they are logged in to Open WebUI and have an active Web browser session. However, if the user has lost their password and cannot log in to Open WebUI, the password must be reset by an SUSE Rancher Prime: RKE2 administrator using the kubectl command.

  1. Generate a new password hash using the htpasswd command. htpasswd is part of the apache2-utils package. 

    > htpasswd -bnBC 10 "" NEW_PASSWORD | tr -d ':\n'

  2. Identify the Pod where the Open WebUI application is running. 

    > kubectl get pod -n suse-private-ai | grep open-webui | grep -v pipelines

  3. Update the password. 

    > kubectl exec open-webui-0 -n suse-private-ai – \
  python3.11 -c 'import sqlite3; con = sqlite3.connect("/usr/lib/python3.11/site-packages/data/webui.db"); \
  con.cursor().execute("UPDATE auth SET password=\"PASSWORD_HASH\" WHERE email=\"admin@example.org\""); con.commit()'

1.6.2 Audio processing stops after several minutes

 

When the voice recorder is used as described in https://documentation.suse.com/suse-ai/1.0/html/openwebui-using/index.html#openwebui-chat-window-usage, Open WebUI tries to process the audio, but it times out without any warning and no text is generated. The API endpoint /audio/api/v1/transcriptions, which receives the audio to be processed, results in a 504 Gateway Timeout.

1.7 For more information

 

2 Setting the default system prompt

 

In Open WebUI, you can set system prompts in several contexts—for the current chat, for the current user, or for a specific AI model. While the prompt set for the current chat overrides the user and model prompts, the user-specific prompt overrides the model prompt only.

2.1 System prompt for the current chat

 

  1. In the top right of the Open WebUI window, click the Chat controls icon.

  2. Unfold the System Prompt tab and paste your system prompt into the empty input text field.

    Chat-specific system prompt
    Figure 3: Chat-specific system prompt
     

  3. Close the Chat controls window to apply the change. The system prompt that you entered will affect the responses from the AI model during the current chat session.

2.2 System prompt for the current user

 

  1. In the top right of the Open WebUI window, click your avatar icon to open the user menu.

  2. Select Settings to open your preferences and click the General tab.

  3. Paste your system prompt in the System Prompt text field.

    User-specific system prompt
    Figure 4: User-specific system prompt
     

  4. Confirm with Save and close the Settings window. The system prompt that you entered will affect the responses from the AI model during the current chat session.

2.3 System prompt for an AI model

 
Requirements
 

  • You must have Open WebUI administrator privileges to access configuration screens or settings mentioned in this section.

  1. In the Open WebUI interface, click Workspace in the upper part of the left panel.

  2. Select Models in the top menu and scroll to the AI model whose system prompt you want to set. Use the pager at the bottom if the model is not listed on the current screen.

  3. Click the corresponding pen icon at its right to open the dialog for editing the model.

  4. In the Model Params section, specify the system prompt in the System prompt field.

    Model system prompt
    Figure 5: Model system prompt
     

  5. Confirm with Save & Update.

2.3.1 For more information

 

Find more details about system prompts in https://documentation.suse.com/suse-ai/1.0/html/AI-system-prompts/index.html.

3 Setting the default AI model

 

In Open WebUI, you can specify an AI model in several contexts—for the current chat, for the current user or for the system default model. While the AI model set for the current chat overrides the user and system model, the user-specific model overrides the system model only.

3.1 AI model for the current chat

 

This topic explains how to specify an AI model for the current chat session.

  1. In the top left of the Open WebUI chat window, click either the currently selected model name or the message saying that no model is specified.

  2. From the drop-down list, select your preferred AI model for the current chat session.

    Chat-specific AI model
    Figure 6: Chat-specific AI model
     
    Tip
    Tip: Your default model

    To make the selected model your default one, click Set as default next to the AI model name.

    Tip
    Tip: Multiple AI modules

    You can select multiple AI modules by clicking the plus sign next to the name of the already selected module.

  3. Your prompts will now be processed by the selected AI model.

3.2 AI model for the current user

 

This topic explains how to specify the default AI model for the current user.

  1. In the top right of the Open WebUI window, click your avatar icon to open the user menu.

  2. Select Settings to open your preferences and click the Interface tab.

  3. From the drop-down list, select your default AI model.

    Default user AI model
    Figure 7: Default user AI model
     

  4. Confirm with Save and close the Settings window. Your prompts will now be processed by the selected AI model unless overridden by selecting a different AI model for the current chat.

3.3 Default system AI model

 

This topic explains how to set the default AI model for all users. The model is then used for new chats if not overridden by specifying a model for a user or a chat.

Requirements
 

  • You must have Open WebUI administrator privileges to access configuration screens or settings mentioned in this section.

  1. In the top right of the Open WebUI window, click your avatar icon to open the user menu and select Admin Panel.

  2. Click Settings in the top left of the Admin Panel and select Users from the left menu.

    Default system AI model
    Figure 8: Default system AI model
     

  3. In the Manage Models section, select the default system AI model. Confirm with Save.

4 Managing custom AI models

 

In Open WebUI, you can create custom AI models based on existing models. Later on, you can modify its settings, or share, clone or delete it.

4.1 Creating custom models

 

This procedure shows how to create custom AI models using Open WebUI.

Requirements
 

  • You must have Open WebUI administrator privileges to access configuration screens or settings mentioned in this section.

  1. In the Open WebUI interface, click Workspace in the upper part of the left panel.

  2. Select Models in the top menu and click the Create a model link under the search bar.

    Managing AI models
    Figure 9: Managing AI models
     

  3. By clicking the big + sign at the top of the form, you can upload an avatar image for your model.

  4. Enter the Name of your new model and change the Model ID if you are not happy with the provided one.

  5. Specify the base AI model on which your model will be based on.

  6. (Optional) Enter a short description of your AI model.

    Avatar, name and description of a new AI model
    Figure 10: Avatar, name and description of a new AI model
     

  7. (Optional) To direct the AI model's responses, specify a System Prompt. Find more details about system prompts in https://documentation.suse.com/suse-ai/1.0/html/AI-system-prompts/index.html.

    Tip
    Tip

    By clicking the small Show option below the system prompt field, you can fine-tune the model's behavior by modifying its advanced settings.

  8. (Optional) Specify one or more Prompt suggestions. They will be shown on the model's chat welcome screen.

    System prompt and prompt suggestions
    Figure 11: System prompt and prompt suggestions
     

  9. (Optional) To extend the model's knowledge base, select one or more documents or collections of documents from the preconfigured list.

    Extending the model's knowledge base
    Figure 12: Extending the model's knowledge base
     

  10. (Optional) In the Tools, Filters, Actions and Capabilities sections, activate the items that you want your model to use.

  11. Describe the model with custom tags so that other users can identify it easily.

    Capabilities and tags
    Figure 13: Capabilities and tags
     
    Tip
    Tip

    To view the JSON representation of your model, click the small Show option to the right of the JSON Preview label.

    Save your changes with the Save & Create button.

4.2 Modifying existing models

 

This procedure shows how to modify an existing custom AI model using Open WebUI.

Requirements
 

  • You must have Open WebUI administrator privileges to access configuration screens or settings mentioned in this section.

  1. In the Open WebUI interface, click Workspace at the top of the left panel and select Models.

  2. In the list of AI models, hover the mouse over the model that you want to modify and click the pen icon on its right.

  3. On the model editing screen, you can modify the settings that are described in Section 4.1, “Creating custom models”.

  4. Save your changes with the Save & Create button.

4.3 Cloning existing models

 

This procedure shows how to clone an existing custom AI model using Open WebUI.

Requirements
 

  • You must have Open WebUI administrator privileges to access configuration screens or settings mentioned in this section.

  1. In the Open WebUI interface, click Workspace at the top of the left panel and select Models.

  2. In the list of AI models, hover the mouse over the model that you want to clone and click the three dots on its right.

  3. From the contextual menu, select Clone to open the same form as described in Section 4.1, “Creating custom models” with all settings copied from the original model. The Name field has (Clone) appended.

  4. Review the settings and create the cloned AI model with Save & Create.

4.4 Importing exported models

 

This procedure shows how to import AI models using Open WebUI.

Requirements
 

  • You must have Open WebUI administrator privileges to access configuration screens or settings mentioned in this section.

  1. In the Open WebUI interface, click Workspace at the top of the left panel and select Models.

  2. Click the Import Models button at the bottom of the screen.

  3. In the file selector, specify the path to the JSON file with the exported models' definitions. Confirm with Open.

    Important
    Important: Models with identical IDs

    AI models with IDs identical to existing models will not be imported.

4.5 Exporting existing models

 

This procedure shows how to export an existing custom AI model using Open WebUI.

Requirements
 

  • You must have Open WebUI administrator privileges to access configuration screens or settings mentioned in this section.

  1. In the Open WebUI interface, click Workspace at the top of the left panel and select Models.

  2. In the list of AI models, hover the mouse over the model that you want to export and click the three dots on its right.

  3. From the contextual menu, select Export.

  4. In the file selector, specify the path where you want to save the exported JSON file. Confirm with Save.

4.6 Exporting all existing models

 

This procedure shows how to export all existing AI models using Open WebUI.

Requirements
 

  • You must have Open WebUI administrator privileges to access configuration screens or settings mentioned in this section.

  1. In the Open WebUI interface, click Workspace at the top of the left panel and select Models.

  2. Click the Export Models button at the bottom of the screen.

  3. In the file selector, specify the path where you want to save the exported JSON file. Confirm with Save.

4.7 Deleting existing models

 

This procedure shows how to delete an existing custom AI model using Open WebUI.

Using the Open WebUI interface, you can delete previously created AI models.

Requirements
 

  • You must have Open WebUI administrator privileges to access configuration screens or settings mentioned in this section.

  1. In the Open WebUI interface, click Workspace at the top of the left panel and select Models.

  2. In the list of AI models, hover the mouse over the model that you want to delete and click the three dots on its right.

  3. From the contextual menu, select Delete and then apply with Confirm.

5 Managing user accounts

 

Open WebUI provides role-based access control (RBAC). Each user is assigned one of the available roles. The role defines whether the user has privileges to perform administration tasks.

5.1 User roles in Open WebUI

 

Open WebUI offers three user roles: “pending,” “user” and “admin.” The following sections provide a detailed description of each role.

5.1.1 The “pending” role

 

Pending users can log in but can only view the Account Activation Pending screen with the administrator name and e-mail. This is the default role for a newly created user account before the administrator accepts it and assigns it the “user” role.

Pending user screen
Figure 14: Pending user screen
 

5.1.2 The “user” role

 

Users with this role can perform the following tasks:

Chat permissions

  • Create new chats with the preconfigured models.

  • View own chats, including archived chats.

  • Pin, rename, clone, archive or unarchive, share, delete, add tags to, and remove tags from existing chats.

  • Generate an overview of a chat, view artifacts, copy content and download chats.

  • Copy, edit and delete their own messages in a chat.

  • Edit, copy and read aloud the responses. View response generation information. Generate images. Mark answers as good or bad. Continue responding and regenerate model messages in a chat.

Current chat controls

  • Modify valves from the tools and functions available on the system.

  • Define the system prompt to be used by the current chat.

  • Modify advanced chat-specific generation parameters.

Current user settings

  • Select theme and language. Activate or deactivate notifications. Define the default system prompt and change advanced generation parameters.

  • Set the default model. Set the user interface, chat and audio related preferences.

  • Personalize the interaction with the AI model by adding “memory”—personal details about yourself—to make responses tailored to you.

  • Set audio preferences.

  • Import and export chats. Archive and delete all user chats.

  • Change your personal details such as name, password and profile image. View JWT token and create new API key.

  • View the installed Open WebUI and Ollama version.

5.1.3 The “admin” role

 

Administrators have full access to Open WebUI resources and can control user management and systems settings. The first Open WebUI account has the “admin” role. Subsequent administrators get their role from existing administrators.

Administrator's task and settings are included either on the Admin Panel or in the Workspace.

Administrators have the permissions that the “user” role has with the following additional privileges.

Admin panel › Dashboard

  • Add new users.

  • View all users and their role, name, e-mail, OAuth 2.0 ID, last active date and the user creation date.

  • Edit name, e-mail and password and delete all users, except for other administrators.

  • View and delete chats of all users, except for other administrators.

  • Change the role of all users, except their own and the first administrator.

Admin panel › Settings

  • Enable or disable new sign-ups. Set the default user role. Enable or disable viewing the administrator details on the Account Activation Pending screen. Enable or disable chat sharing and message rating. Set the JWT expiration and webhook URL.

  • Enable or disable chat deletion, chat editing and temporary chat. Set the default model and specify allowed models.

  • View, change and add OpenAI API URL and Ollama API URL.

  • Update all models, pull models from ollama.com, delete, create and upload models.

  • Change documents settings, set the RAG template, and reset document storage.

  • Enable or disable Web search, change the Web search engine, and set Web loader settings.

  • Set the default model used for the title generation. Change the default prompt suggestions for the New Chat screen.

  • Change audio settings.

  • Enable or disable image generation and change image generation settings.

  • Change pipeline settings.

  • Import or export configuration files, download database, and export all chats from all users.

Workspace › Models

Manage models. Add models by pulling from ollama.com or by creating custom models from base models. Change order of models or hide models from the model list.

Workspace › Knowledge

Manage collections of documents. Add documents to collections.

Workspace › Prompts

Manage system prompts. Share prompts, import and export all prompts.

Workspace › Tools

Manage tools. Import and export tools.

Workspace › Functions

Manage functions. Enable or disable functions. Globally enable or disable functions. Import and export functions. Change valves settings.

5.1.4 Creating user accounts

 

This topic explains how to create a new user account with an assigned role in Open WebUI. It also describes how to create multiple user accounts from a CSV list.

Procedure 1: Creating a single user account
 
Requirements
 

  • You must have Open WebUI administrator privileges to access configuration screens or settings mentioned in this section.

  1. Click your avatar button at the top right of the chat screen and select Admin Panel.

  2. On the Dashboard screen, click the “+” sign at the top right.

    Creating new users
    Figure 15: Creating new users
     

  3. On the Form screen, select the user role and enter the user name, e-mail and password.

  4. Confirm with Submit.

Procedure 2: Creating multiple user accounts from a CSV list
 
Requirements
 

  • You must have Open WebUI administrator privileges to access configuration screens or settings mentioned in this section.

  1. Click your avatar button at the top right of the chat screen and select Admin Panel.

  2. On the Dashboard screen, click the “+” sign at the top right.

  3. On the CSV Import screen, specify the path to your CSV file with the list of user accounts to be created.

    Creating multiple users
    Figure 16: Creating multiple users
     
    Tip
    Tip

    The CSV list needs to have the four fields Name,Email,Password,Role per line separated by commas, for example: 

    Example User,tux@example.com,SECRET_PASSWD,user

  4. Confirm with Submit.

5.1.5 Assigning roles to existing users

 

This topic explains how to assign one of the available roles to an existing user account. To assign the role when creating a new user account, refer to Section 5.1.4, “Creating user accounts”.

Requirements
 

  • You must have Open WebUI administrator privileges to access configuration screens or settings mentioned in this section.

  1. Click your avatar button at the top right of the chat screen and select Admin Panel.

  2. On the Dashboard screen, scroll to the user whose role you want to modify. Use the pager at the bottom if the user is not listed on the current screen.

    Assigning user role
    Figure 17: Assigning user role
     

  3. Click the role button to the left of the user name to cycle through “pending” → “user” → “admin” role.

5.1.6 Modifying existing user accounts

 

This topic explains how to modify existing user accounts in Open WebUI.

Requirements
 

  • You must have Open WebUI administrator privileges to access configuration screens or settings mentioned in this section.

  1. Click your avatar button at the top right of the chat screen and select Admin Panel.

  2. On the Dashboard screen, scroll to the user whose account you want to modify. Use the pager at the bottom if the user is not listed on the current screen.

  3. Click the corresponding pen icon at their right to open the Edit User dialog.

    Modifying user accounts
    Figure 18: Modifying user accounts
     

  4. Modify the user's e-mail, name or password and confirm with Submit.

5.1.7 Revoking access to Open WebUI

 

This topic explains how to revoke access to Open WebUI for an existing user account.

Requirements
 

  • You must have Open WebUI administrator privileges to access configuration screens or settings mentioned in this section.

  1. Click your avatar button at the top right of the chat screen and select Admin Panel.

  2. On the Dashboard screen, scroll to the user whose access to Open WebUI you want to revoke. Use the pager at the bottom if the user is not listed on the current screen.

    1. To revoke the access while maintaining the user account, click the user's role button to the left of their name until it becomes “pending.”

    2. To delete the user account from the system, click the trash icon on the right and confirm the deletion.

6 Managing a knowledge base

 

Knowledge base is a custom collection of resources on a specific topic. It can include documentation, tutorials, reference guides or short text snippets. Knowledge base collections extend the data on which the selected AI model is trained to make its context more specific.

6.1 Creating a knowledge base

 

This topic explains how to create a new knowledge base to extend the AI model's context on a specific topic.

Requirements
 

  • You must have Open WebUI administrator privileges to access configuration screens or settings mentioned in this section.

  1. In the Open WebUI interface, click Workspace in the upper part of the left panel.

  2. Select Knowledge in the top menu and click the “plus” sign in the right.

  3. Enter a title and a description for the knowledge base and confirm with Create Knowledge.

    Creating a new knowledge base
    Figure 19: Creating a new knowledge base
     

  4. (Optional) Add content to the knowledge base as described in Section 6.3, “Adding content to a knowledge base”.

Tip
Tip

You can access knowledge base collections from the Open WebUI chat input field by pressing the # character as described in https://documentation.suse.com/suse-ai/1.0/html/openwebui-using/index.html#openwebui-input-field-collections.

6.2 Deleting a knowledge base

 

This topic explains how to delete an existing knowledge base.

Requirements
 

  • You must have Open WebUI administrator privileges to access configuration screens or settings mentioned in this section.

  1. In the Open WebUI interface, click Workspace in the upper part of the left panel.

  2. Select Knowledge in the top menu and scroll to the knowledge base that you want to delete. Use the pager at the bottom if the knowledge base is not listed on the current screen.

  3. In the knowledge base tile's top right corner, click the three dots and select Delete. Confirm the deletion.

6.3 Adding content to a knowledge base

 

This topic explains how to add a new content to an existing knowledge base.

Requirements
 

  • You must have Open WebUI administrator privileges to access configuration screens or settings mentioned in this section.

  1. In the Open WebUI interface, click Workspace in the upper part of the left panel and scroll to the knowledge base that you want to add content to. Use the pager at the bottom if the knowledge base is not listed on the current screen.

  2. Click the knowledge base name and in the Search Collection files, click the “plus” sign.

  3. From the drop-down list, you can select one of the following actions:

    Upload files

    Select one or more documents to be added to the collection.

    Upload directory

    Select a directory whose contents will be added to the collection one item at a time.

    Warning
    Warning: Wait until the processing is finished

    Do not leave the collection's screen while the selected files are being uploaded and processed. Leaving the screen may disrupt the file processing.

    Sync directory

    The selected directory is synchronized with the content of the collection.

    Warning
    Warning: The collection content will be deleted

    Before Open WebUI starts to synchronize the directory, all existing resources in the collection are deleted.

    Warning
    Warning: Wait until the processing is finished

    Do not leave the collection's screen while the selected files are being uploaded and processed. Leaving the screen may disrupt the file processing.

    Add text entry

    You can add a text snippet written in a Markdown format to the collection directly.

    Adding a text entry into Open WebUI collection
    Figure 20: Adding a text entry into Open WebUI collection
     

    Confirm with Upload.

  4. The uploaded resource is processed and analyzed so that you can use it as a context for selected AI models. The processing time may vary depending on the size of the uploaded resources.

    After a resource is added, you can view and edit its extracted textual information by clicking its name.

    Viewing an extracted resource
    Figure 21: Viewing an extracted resource
     

6.4 Removing content from a knowledge base

 

This topic explains how to remove an existing content from a knowledge base.

Requirements
 

  • You must have Open WebUI administrator privileges to access configuration screens or settings mentioned in this section.

  1. In the Open WebUI interface, click Workspace in the upper part of the left panel and scroll to the knowledge base that you want to add content to. Use the pager at the bottom if the knowledge base is not listed on the current screen.

  2. Click the knowledge base name to view all the resources in the collection.

    Collection resources
    Figure 22: Collection resources
     

  3. Hover over the resource's name and click the little cross sign at its top right. The resource will be deleted and its processed data removed from the collection.

7 Managing system prompts

 

System prompts are a powerful core component that shapes AI model behavior. They provide precise control over how AI models process and respond to user inputs.

Find more details about system prompts in Guiding the AI Model with System Prompts.

7.1 Creating a system prompt

 

This procedure shows how to create a new system prompt.

Requirements
 

  • You must have Open WebUI administrator privileges to access configuration screens or settings mentioned in this section.

  1. In the Open WebUI interface, click Workspace in the upper part of the left panel.

  2. Select Prompts in the top menu and click the “plus” sign on the right.

    Creating a new system prompt
    Figure 23: Creating a new system prompt
     

  3. Enter the title and content of the new system prompt, and optionally, update the command that invokes the prompt from the chat input field.

  4. Confirm with Save & Create.

Tip
Tip

You can access preconfigured system prompts from the Open WebUI chat input field by pressing the / character as described in https://documentation.suse.com/suse-ai/1.0/html/openwebui-using/index.html#openwebui-input-field-specify-system-prompt.

7.2 Modifying existing system prompts

 

This procedure shows how to modify an existing system prompt.

Requirements
 

  • You must have Open WebUI administrator privileges to access configuration screens or settings mentioned in this section.

  1. In the Open WebUI interface, click Workspace in the upper part of the left panel and select Prompts.

  2. Scroll to the prompt that you want to modify. Use the pager at the bottom if the prompt is not listed on the current screen.

  3. Click the corresponding pen icon on the right to open the form that enables you to edit the title and the description of the prompt.

  4. Confirm with Save & Update.

7.3 Cloning existing system prompts

 

This procedure shows how to clone an existing system prompt.

Requirements
 

  • You must have Open WebUI administrator privileges to access configuration screens or settings mentioned in this section.

  1. In the Open WebUI interface, click Workspace in the upper part of the left panel and select Prompts.

  2. Scroll to the prompt that you want to clone. Use the pager at the bottom if the prompt is not listed on the current screen.

  3. Click the three dots to the right of the prompt that you want to clone and select Clone.

    Cloning a system prompt
    Figure 24: Cloning a system prompt
     

  4. Optionally update the title, the content and the activating command of the prompt. The command must be unique among all other existing system prompts.

  5. Confirm with Save & Create.

7.4 Importing exported system prompts

 

This procedure shows how to import a system prompt that was previously exported.

Requirements
 

  • You must have Open WebUI administrator privileges to access configuration screens or settings mentioned in this section.

  1. In the Open WebUI interface, click Workspace in the upper part of the left panel and select Prompts.

  2. Click Import Prompts in the bottom right of the prompt list.

  3. Navigate to the JSON file with previously exported prompts and confirm with Open.

    Warning
    Warning: Unique command

    The activating command of the imported prompts must be unique among existing prompts. Prompts with an existing activating command will not be imported.

7.5 Exporting existing system prompts

 

This procedure shows how to export an existing system prompt.

Requirements
 

  • You must have Open WebUI administrator privileges to access configuration screens or settings mentioned in this section.

  1. In the Open WebUI interface, click Workspace in the upper part of the left panel and select Prompts.

  2. Scroll to the prompt that you want to export. Use the pager at the bottom if the prompt is not listed on the current screen.

  3. Click the three dots to the right of the prompt that you want to clone and select Export.

  4. Navigate to the path where you want to save the JSON file with the exported prompt. Confirm with Save.

7.6 Exporting all existing system prompts

 

This procedure shows how to export all existing system prompts.

Requirements
 

  • You must have Open WebUI administrator privileges to access configuration screens or settings mentioned in this section.

  1. In the Open WebUI interface, click Workspace in the upper part of the left panel and select Prompts.

  2. Click Export Prompts in the bottom right of the prompt list.

  3. Enter the path to the JSON file that will contain the exported prompts and confirm with Save.

7.7 Deleting existing system prompts

 

This procedure shows how to delete an existing system prompt.

Requirements
 

  • You must have Open WebUI administrator privileges to access configuration screens or settings mentioned in this section.

  1. In the Open WebUI interface, click Workspace in the upper part of the left panel and select Prompts.

  2. Scroll to the prompt that you want to clone. Use the pager at the bottom if the prompt is not listed on the current screen.

  3. Click the three dots to the right of the prompt that you want to clone and select Delete. Confirm the deletion.

8 Authenticating via Okta

 

Okta is an identity management service that allows you to manage users and user access to applications. This topic describes how to use Okta as the identity manager for Open WebUI.

The process of enabling Okta as the identity manager for Open WebUI consists of several steps:

  1. Configuring Okta

  2. Gathering data for Open WebUI configuration

  3. Configuring Open WebUI for Okta integration

Tip
Tip

This example procedure shows the minimal configuration required to use Okta with Open WebUI. Customize the configuration to your specific scenario and refer to the Okta documentation at https://help.okta.com/en-us/content/index.htm?cshid=csh-index for more information.

Procedure 3: Configuring Okta
 
Important
Important

You must have administrator privileges to perform the following steps.

  1. Log in to Okta and access the Admin panel.

    Okta administrator panel
    Figure 25: Okta administrator panel
     

  2. Go to Applications › Create App Integration.

  3. In the Sign-in method section, select OIDC - OpenID Connect. In the Application type section, select Web Application. Click Next to proceed.

    Okta configuration window
    Figure 26: Okta configuration window
     

  4. In the General Settings section, set the App integration name to Open WebUI. In the Grant type section, toggle the Refresh token option. Optionally, specify an image for the application logo.

  5. Set Sign-in redirect URIs to the callback URI, for example, http://localhost:8080/oauth/oidc/callback. The URI follows this pattern: 

    PROTOCOL://OPENWEBUI_DOMAIN:OPTIONAL_PORT_NUMBER/oauth/oidc/callback

  6. Similarly, set Sign-out redirect URIs to 

    PROTOCOL://OPENWEBUI_DOMAIN:OPTIONAL_PORT_NUMBER

  7. In the Assignments section, toggle the Skip group assignment for now option. Save changes.

    Tip
    Tip: Open WebUI application settings

    After the integration is created, you can access the Open WebUI application settings by clicking the related hyperlink. It is required for final configurations and for retrieving the necessary information for the Open WebUI configuration further on.

    Open WebUI application access in Okta
    Figure 27: Open WebUI application access in Okta
     

  8. In the General Settings section, click Edit and in the Login section, set the Sign-in redirect URIs option to the URI in this format: 

    PROTOCOL://OPENWEBUI_DOMAIN:OPTIONAL_PORT_NUMBER/oauth/oidc/callback

  9. In the Login initiated by section, select App Only. Confirm with Save.

Procedure 4: Gathering data for Open WebUI configuration
 

  1. In the top-right menu, copy the Okta domain. In this example, it is my-domain.okta.com.

    Okta domain name
    Figure 28: Okta domain name
     

  2. Go to Applications › Open WebUI and retrieve the Client ID and Client Secret.

    Client ID and secret in Okta
    Figure 29: Client ID and secret in Okta
     
    Tip
    Tip

    The previously gathered values are used by Open WebUI via the following environment variables as described in Procedure 5, “Configuring Open WebUI for Okta integration”. 

    OPENID_PROVIDER_URL="OKTA_DOMAIN/.well-known/openid-configuration"
OAUTH_CLIENT_ID="CLIENT_ID"
OAUTH_CLIENT_SECRET="CLIENT_SECRET"
Procedure 5: Configuring Open WebUI for Okta integration
 

To configure Open WebUI to use Okta as an authentication provider, you must set up specific environment variables. For more information about OAuth 2.0 in Open WebUI and a complete description of each environment variable, refer to Open WebUI SSO documentation.

  1. Add the following environment variables to your Helm charts values.yaml file in the extraEnvVars section. 

    ENABLE_OAUTH_SIGNUP=true 1
OAUTH_MERGE_ACCOUNTS_BY_EMAIL=true 2
OAUTH_CLIENT_ID=CLIENT_ID 3
OAUTH_CLIENT_SECRET=CLIENT_SECRET
OPENID_PROVIDER_URL=OKTA_DOMAIN/.well-known/openid-configuration
OAUTH_PROVIDER_NAME=Okta
OAUTH_SCOPES=openid email profile 4

    1

    Allow account creation when logging in with OAuth 2.0.

    2

    Allow logging into an account that matches the e-mail address provided by the OAuth 2.0 provider (optional).

    3

    Variable set for the authentication provider gathered from Okta.

    4

    Specify scopes to request (optional, defaults to openid email profile).

    The code block should look as follows: 

    extraEnvVars:
  - name: ENABLE_OAUTH_SIGNUP
    value: "True"
  - name: OAUTH_MERGE_ACCOUNTS_BY_EMAIL
    value: "True"
  - name: OAUTH_CLIENT_ID
    value: "CLIENT_ID"
  - name: OAUTH_CLIENT_SECRET
    value: "CLIENT_SECRET"
  - name: OPENID_PROVIDER_URL
    value: "OKTA_DOMAIN/.well-known/openid-configuration"
  - name: OAUTH_PROVIDER_NAME
    value: "Okta"
  - name: OAUTH_SCOPES
    value: "openid email profile"

  2. Restart Open WebUI to apply the changes.

  3. If the configuration went well, you should see the Continue with Okta button as an option on both sign-in and sign-up pages. After you click the button, you will be redirected to the Okta login page, or automatically signed in if you are already logged in to Okta.

    Open WebUI
    Figure 30: Open WebUI