Open Chat Playground

Open Chat Playground (OCP) is a web UI that is able to connect virtually any LLM from any platform.

Supported platforms

• [x] Amazon Bedrock
• [x] Azure AI Foundry
• [x] GitHub Models
• [x] Google Vertex AI
• [x] Docker Model Runner
• [x] Foundry Local
• [x] Hugging Face
• [x] Ollama
• [x] Anthropic
• [ ] ~~Naver~~
• [x] LG
• [x] OpenAI
• [x] Upstage

Prerequisites

• Azure Subscription
• .NET SDK 9
• Visual Studio Code + C# DevKit or Visual Studio 2022 v17.14+
• Docker Desktop
• Azure Developer CLI
• GitHub CLI
• PowerShell 7.5+ 👉 Windows only

Getting started

Get the repository ready

1. Login to GitHub.

   gh auth login
   

2. Check login status.

   gh auth status
   

3. Fork this repository to your account and clone the forked repository to your local machine.

   gh repo fork aliencube/open-chat-playground --clone --default-branch-only
   

4. Navigate to the cloned repository.

   cd open-chat-playground
   

Run on local machine

• Use Amazon Bedrock
• Use Azure AI Foundry
• Use GitHub Models
• Use Google Vertex AI
• Use Docker Model Runner
• Use Foundry Local
• Use Hugging Face
• Use Ollama
• Use Anthropic
• Use LG
• Use OpenAI
• Use Upstage

Run in local container

• Use Amazon Bedrock
• Use Azure AI Foundry
• Use GitHub Models
• Use Google Vertex AI
• Use Docker Model Runner
• ~~Use Foundry Local~~ 👉 NOT SUPPORTED
• Use Hugging Face
• Use Ollama
• Use Anthropic
• Use LG
• ~~Use Naver~~ 👉 NOT SUPPORTED
• Use OpenAI
• Use Upstage

Run on Azure

• Use Amazon Bedrock
• Use Azure AI Foundry
• Use GitHub Models
• Use Google Vertex AI
• ~~Use Docker Model Runner~~ 👉 NOT SUPPORTED
• ~~Use Foundry Local~~ 👉 NOT SUPPORTED
• Use Hugging Face
• Use Ollama
• Use Anthropic
• Use LG
• ~~Use Naver~~ 👉 NOT SUPPORTED
• Use OpenAI
• Use Upstage

Run as API

1. Run the app - either locally or remotely.

2. Navigate to {{APP_URL}}/openapi.json to see the OpenAPI doc. NOTE: the OpenAPI doc is only able to see on a local machine, not on a remote instance.

3. Send a POST request to {{APP_URL}}/api/chat/responses with a request payload like:

   [
     {
       "role": "system",
       "message": "You're a friendly assistant"
     },
     {
       "role": "user",
       "message": "Why is the sky blue?"
     }
   ]
   

4. Confirm the response payload like:

   [
     {
       "role": "assistant",
       "message": "Because "
     },
     {
       "role": "assistant",
       "message": "It's "
     },
     {
       "role": "assistant",
       "message": "... "
     },
     ...
     {
       "role": "assistant",
       "message": "... "
     }
   ]
   

Run tests

Build app

1. Make sure you are at the repository root.

   cd $REPOSITORY_ROOT
   

2. Build the app.

   dotnet restore && dotnet build
   

Unit tests

1. Make sure you are at the repository root.

   cd $REPOSITORY_ROOT
   

2. Run tests.

   dotnet test --filter "Category=UnitTest"
   

Integration tests

1. Make sure you have the Ollama server up and running on your local machine.

   ollama serve
   ollama run hf.co/Qwen/Qwen3-0.6B-GGUF
   

2. Make sure you are at the repository root.

   cd $REPOSITORY_ROOT
   

3. Install playwright.

   pwsh $REPOSITORY_ROOT/test/OpenChat.PlaygroundApp.Tests/bin/Debug/net{YOUR_VERSION}/playwright.ps1 install
   

4. Run the app.

   dotnet run --project $REPOSITORY_ROOT/src/OpenChat.PlaygroundApp -- --connector-type HuggingFace
   

5. Run tests.

   # With LLM provider
   dotnet test --filter "Category=IntegrationTest"
   

   # Without LLM provider
   dotnet test --filter "Category=IntegrationTest & Category!=LLMRequired"
   

Configure GitHub Actions for CI/CD Pipeline

1. Make sure you are at the repository root.

   cd $REPOSITORY_ROOT
   

2. Make sure you've logged in to Azure.

   azd auth login --check-status
   

3. Run pipeline config.

   azd pipeline config
   

4. Answer the question below:

   • ? Select how to authenticate the pipeline to Azure 👉 Federated Service Principal (SP + OIDC)
   • ? Would you like to commit and push your local changes to start the configured CI pipeline? 👉 No

5. Once the configuration is done, push a new commit to GitHub to run the GitHub Actions workflow.