Skip to content

Latest commit

 

History

History
194 lines (146 loc) · 44.1 KB

README.md

File metadata and controls

194 lines (146 loc) · 44.1 KB

DIAL Chat

This is a default UI for AI DIAL. AI DIAL can be used as headless system, but UI is recommended to learn the capability.

Originally forked from chatbot-ui and then completely reworked and published under apache 2.0 license, while code taken from original repository is still subject to original MIT license. Due to rework we introduced lots of new features such as various IDP support, model side-by-side compare, DIAL extensions support, conversation replays, branding and many more.

ai-dial-chat

Docs

  • AI DIAL Overlay documentation is placed here.
  • Theming and branding documentation is placed here.
  • In AI DIAL repository, you can find a User Guide for the AI DIAL Chat.

Overview

This project is a web application built using Next.js, a React framework with server-rendering capabilities. It can be easily customized and adapted to your needs by following the steps mentioned below.

Helm Deployment

You have the option to deploy the chat application on the Kubernetes cluster by utilizing an umbrella dial Helm chart, which also deploys other AI DIAL components. Alternatively, you can establish a dedicated Helm chart exclusively for the chat application using the dial-extension Helm chart.

Refer to Examples for guidelines.

In any case, in your Helm chart values.yaml file, it is necessary to provide application's configurations described in the Environment Variables section.

Set Up Developer Environment

This project is managed by NX.

Before diving into the development environment, install Node.js (version >=18.18.2) and npm (version >=9.8.1) on your system. If you don't already have them, follow the instruction.

Once you have Node.js and npm installed, follow these steps to set up your developer environment:

  1. Clone the AI DIAL Chat repository:
git clone https://github.com/epam/ai-dial-chat.git
  1. Install project dependencies:
npm install
  1. Create .env.local file in the \apps\chat project directory and add the required variables with appropriate values. These three are the only required environment variables. Refer to Environment Variables to learn more.
DIAL_API_HOST="ADD_VALUE_HERE"
DIAL_API_KEY="ADD_VALUE_HERE"
NEXTAUTH_SECRET="ADD_VALUE_HERE"
  1. To start the development server, run:
npm run nx serve chat

Once the server is up and running, open http://localhost:3000 in your browser to view the AI DIAL Chat application.

To run the optimized production build, execute this command:

npm run nx serve chat --configuration=production

This will start a production server on the default port 3000.

Build

To create an optimized build of your application, run the following command:

npm run nx build chat --configuration=production

After running the command, you will see a .next folder created in your project directory with the optimized output.

Test

To run the unit tests suite for your application, execute the following command:

npm run nx test chat

To run e2e tests, run the following command:

npm run nx e2e chat-e2e

For more information refer to E2E tests documentation.

Environment Variables

AI DIAL Chat uses environment variables for configuration. All environment variables that can be used to configure settings and behavior of the application are included in the .env file.

Selected variables were predefined for the development purposes in the .env.development file.

Variable Required Description Available Values Default values
DIAL_API_HOST Yes AI DIAL Core API Host.
Refer to AI DIAL Core.
Any string
DIAL_API_KEY Yes AI DIAL Core API Key.
Refer to AI DIAL Core to learn how to set up AI DIAL Core and define API keys.
Any string
DIAL_API_VERSION No AI DIAL API Version Any string 2024-02-01
APP_BASE_PATH No The root directory in the file system where the application is located Any string
APP_BASE_ORIGIN Optional A base URL or origin of the application.
Required if APP_BASE_PATH is set.
Any string
ALLOWED_IFRAME_ORIGINS No Specifies all URLs that are allowed to embed AI DIAl Chat within an iframe.
Necessary for DIAL Overlay.
Any origin valid format. List of space separated URLs. none
ALLOWED_IFRAME_SOURCES No iFrame sources (e.g. visualizers) that are allowed to load content into an iframe in the AI DIAL Chat application Any valid source format. List of space separated sources. none
IS_IFRAME No Is iFrame flag enables/disables the DIAL Overlay true, false false
ENABLED_FEATURES No Features enabled in the AI DIAL Chat application Refer to Features to view all the available features.
PUBLICATION_FILTERS No Defines types of publications that can be retrieved from the database Any string. Values must be separated by a comma. title,role
ADMIN_ROLE_NAMES No Defines the name of the administrator role Any string. Values must be separated by a comma. admin
DIAL_ROLES_FIELD No Defines the path of the roles field in JWT token Any string. Value can be dot-separated. E.g. path realm_access.roles if there is a claim realm_access: { roles: ['admin'] } dial_roles
NEXT_PUBLIC_APP_NAME No Application name Any string AI Dial
NEXT_PUBLIC_DEFAULT_SYSTEM_PROMPT No Default system prompt Any string
NEXT_PUBLIC_DEFAULT_TEMPERATURE No Default temperature 0 to 1
DEFAULT_MODEL No A model that is pre-selected for a new conversation.
If not defined, the recently-used model or (if not available) a model from the local storage will be used. If none of the above is available - the first model from the AI DIAL Core config will be used as a pre-selected conversation model.
Any string First available model from AI DIAL Core config listing
NEXT_PUBLIC_DEFAULT_ASSISTANT_SUB_MODEL No A default pre-selected model for any assistant Any string gpt-4
RECENT_MODELS_IDS No A list of IDs for recently-used models Any string
RECENT_ADDONS_IDS No A list of IDs for recently-used addons Any string
THEMES_CONFIG_HOST No The host URL for a custom themes configuration.
Can lead to public and private resource.
Refer to Theme Customization to learn more.
Any string Dark
FOOTER_HTML_MESSAGE No A message that will be displayed in the application's footer.
Specify in HTML format.
Any string
ANNOUNCEMENT_HTML_MESSAGE No A message of the announcement banner.
Specify in HTML format.
Any string
AZURE_FUNCTIONS_API_HOST No Azure Functions API Host Any string
REPORT_ISSUE_CODE No Report Issue Code used when sending report issue info to Azure Functions API Host Any string
REQUEST_API_KEY_CODE No Request API Key Code used when sending request api key info to Azure Functions API Host Any string
CODE_GENERATION_WARNING No Warning text regarding code generation Any string
SHOW_TOKEN_SUB No Show token sub in refresh login error logs true, false false
STORAGE_TYPE No Type of storage used for getting and saving information generated by a user.
Note: currently, api is the only supported value.
browserStorage, api api
KEEP_ALIVE_TIMEOUT No Determines the maximum time in milliseconds for an idle connection before it is closed by the server.
This is needed because infrastructures usually have a default keep alive timeout 60 seconds and the Next server should have a larger value.
Note: this variable is used only when running a dockerfile.
Any number string 61000
MAX_PROMPT_TOKENS_DEFAULT_PERCENT No A share in percent of MAX_PROMPT_TOKENS_DEFAULT_VALUE.
If not set, the value of MAX_PROMPT_TOKENS_DEFAULT_VALUE applies.
Any string 75
MAX_PROMPT_TOKENS_DEFAULT_VALUE No The default maximum limit of tokens permissible for any model.
Applies if a limit is not defined in the AI DIAL Core configuration.
Any string 2000
CUSTOM_VISUALIZERS No A list of configurations of custom visualizers.
Refer to Chat Visualizer to learn more.
[{ title, description, icon, contentType, url }]
ALLOW_VISUALIZER_SEND_MESSAGES No Allow the addition of custom buttons or input fields to enable a custom visualizer to send SEND_MESSAGE events on behalf of a user, in the form of messages within a conversation. true, false false

Environment Variables for the Configuration of Opentelemetry

All standard env variables you could find in the official opentelemetry documentation. If no value set for the OTEL_METRICS_EXPORTER then OpenTelemetry Prometheus Metric Exporter will be used. If value set to "otlp" the OpenTelemetry Collector Metrics Exporter for web and node will be used.

Environment Variables for the Configuration of Auth Providers

Refer to IDP Configuration to view configuration examples for supported IDP providers.

The table below presents a list of environment variables you can use to configure a specific IDP provider.

NOTE: to test the AI DIAL Chat application in an unauthenticated mode, do not provide any of these variables. The only required variable to launch the application is NEXTAUTH_SECRET.

Variable Required Description Available Values Default values
NEXTAUTH_URL Optional.
Required for production deployments.
NextAuth URL Any string
NEXTAUTH_SECRET Yes NextAuth Secret (generate by openssl rand -base64 32 for example) Any string
AUTH_AUTH0_AUDIENCE No Auth0 Audience Any string
AUTH_AUTH0_CLIENT_ID No Auth0 Client ID Any string
AUTH_AUTH0_HOST No Auth0 Host Any string
AUTH_AUTH0_NAME No Auth0 Name Any string
AUTH_AUTH0_SECRET No Auth0 Secret Any string
AUTH_AUTH0_SCOPE No Auth0 Scope Any string openid email profile offline_access
AUTH_AZURE_AD_CLIENT_ID No A unique identifier for the client application registered in Azure Active Directory (AD). It is used to authenticate the client application when accessing Azure AD resources. Any string
AUTH_AZURE_AD_NAME No A name of the Azure AD tenant. It is used to specify the specific Azure AD instance to authenticate against. Any string
AUTH_AZURE_AD_SECRET No Also known as the client secret or application secret, this parameter is a confidential string that authenticates and authorizes the client application to access Azure AD resources. It serves as a password for the client application. Any string
AUTH_AZURE_AD_TENANT_ID No Tenant ID refers to a globally unique identifier (GUID) that represents a specific Azure AD tenant. It is used to identify and authenticate the Azure AD tenant that the client application belongs to. Any string
AUTH_AZURE_AD_SCOPE No This parameter specifies the level of access and permissions that the client application requests when making a request to Azure AD resources. It defines the resources and actions that the application can access on behalf of a user or itself. Any string openid profile user.Read email offline_access
AUTH_GITLAB_CLIENT_ID No GitLab Client ID Any string
AUTH_GITLAB_HOST No GitLab Host Any string
AUTH_GITLAB_NAME No GitLab Name Any string
AUTH_GITLAB_SECRET No GitLab Secret Any string
AUTH_GITLAB_SCOPE No GitLab Scope Any string read_user
AUTH_GOOGLE_CLIENT_ID No Google Client ID Any string
AUTH_GOOGLE_NAME No Google Name Any string
AUTH_GOOGLE_SECRET No Google Secret Any string
AUTH_GOOGLE_SCOPE No Google Scope Any string openid email profile offline_access
AUTH_KEYCLOAK_CLIENT_ID No Keycloak Client ID Any string
AUTH_KEYCLOAK_HOST No Keycloak Host Any string
AUTH_KEYCLOAK_NAME No Keycloak Name Any string
AUTH_KEYCLOAK_SECRET No Keycloak Secret Any string
AUTH_KEYCLOAK_SCOPE No Keycloak Scope Any string openid email profile offline_access
AUTH_PING_ID_CLIENT_ID No PingID Client ID Any string
AUTH_PING_ID_HOST No PingID Host Any string
AUTH_PING_ID_NAME No PingID Name Any string
AUTH_PING_ID_SECRET No PingID Secret Any string
AUTH_PING_ID_SCOPE No PingID Scope Any string offline_access
AUTH_COGNITO_CLIENT_ID No Cognito Client ID Any string
AUTH_COGNITO_HOST No Cognito Host Any string
AUTH_COGNITO_NAME No Cognito Name Any string
AUTH_COGNITO_SECRET No Cognito Secret Any string
AUTH_COGNITO_SCOPE No Cognito Scope Any string openid email profile
AUTH_OKTA_CLIENT_ID No Okta Client ID Any string
AUTH_OKTA_CLIENT_SECRET No Okta Client Secret Any string
AUTH_OKTA_ISSUER No Okta domain issuer Any string
AUTH_OKTA_SCOPE No Okta Scope Any string openid email profile