Next Gen UI MCP Server Container

This module is part of the Next Gen UI Agent project.

Next Gen UI Agent MCP Server container image.

Provides

container image to easily run Next Gen UI Agent MCP server

Installation

podman pull quay.io/next-gen-ui/mcp

Usage

Locally

podman run --rm -it -p 5100:5100 --env MCP_PORT="5100" \
       --env NGUI_MODEL="llama3.2" --env NGUI_PROVIDER_API_BASE_URL=http://host.containers.internal:11434 --env NGUI_PROVIDER_API_KEY="ollama" \
       quay.io/next-gen-ui/mcp

Openshift

oc login ...
oc project next-gen-ui # or another
oc apply -f deployment.yaml

Configuration

The MCP server container can be configured via environment variables. All configuration options available to the standalone server are supported. All env variables are mapped to Next Gen UI MCP server documented in the MCP Server Guide.

Environment Variables Reference

Environment Variable	Default Value	Description
`MCP_TRANSPORT`	`streamable-http`	Transport protocol (`stdio`, `sse`, `streamable-http`)
`MCP_HOST`	`0.0.0.0`	Host to bind to (for HTTP transports)
`MCP_PORT`	`5000`	Port to bind to (for HTTP transports)
`MCP_TOOLS`	`all`	List of enabled tools (comma separated)
`MCP_STRUCTURED_OUTPUT_ENABLED`	`true`	Enable or disable structured output
`NGUI_COMPONENT_SYSTEM`	`json`	UI Component system (`json`, `rhds`)
`NGUI_PROVIDER`	`langchain`	Inference provider (`mcp`, `llamastack`, `langchain`)
`NGUI_MODEL`	`gpt-4o`	Model name (required for other than `mcp` providers)
`NGUI_PROVIDER_API_BASE_URL`	-	Base URL for OpenAI-compatible API (if `langchain` is used)
`NGUI_PROVIDER_API_KEY`	-	API key for the LLM provider (if `langchain` is used)
`NGUI_PROVIDER_LLAMA_URL`	-	LlamaStack server URL (if `llamastack` is used)
`NGUI_CONFIG_PATH`	-	Path to Next Gen UI YAML configuration files (comma separated).

Providers

The Next Gen UI MCP server supports three inference providers, controlled by the NGUI_PROVIDER environment variable:

Selects the inference provider to use for generating UI components:

Provider `mcp`

Uses Model Context Protocol sampling to leverage the client's LLM capabilities. No additional configuration required as it uses the connected MCP client's model, but MCP client has to support this feature!

Provider `langchain`:

Uses LangChain inference provider, with OpenAI-compatible APIs.

Requires:

NGUI_MODEL: Model name (e.g., gpt-4o, llama3.2)
NGUI_PROVIDER_API_KEY: API key for the provider
NGUI_PROVIDER_API_BASE_URL (optional): Custom base URL for OpenAI-compatible APIs like Ollama. OpenAI API by default.

Examples:

OpenAI: https://api.openai.com/v1 (default)
Ollama: http://host.containers.internal:11434/v1

Provider `llamastack`:

Uses Remote LlamaStack server inference provider.

Requires:

NGUI_MODEL: Model name available on the LlamaStack server
NGUI_PROVIDER_LLAMA_URL: URL of the LlamaStack server

Usage Examples

Basic Usage with Ollama (Local LLM)

podman run --rm -it -p 5000:5000 \
    --env MCP_PORT="5000" \
    --env NGUI_MODEL="llama3.2" \
    --env NGUI_PROVIDER_API_BASE_URL="http://host.containers.internal:11434/v1" \
    --env NGUI_PROVIDER_API_KEY="ollama" \
    quay.io/next-gen-ui/mcp

OpenAI Configuration

podman run --rm -it -p 5000:5000 \
    --env NGUI_MODEL="gpt-4o" \
    --env NGUI_PROVIDER_API_KEY="your-openai-api-key" \
    quay.io/next-gen-ui/mcp

Remote LlamaStack Server Provider

podman run --rm -it -p 5000:5000 \
    --env NGUI_PROVIDER="llamastack" \
    --env NGUI_MODEL="llama3.2-3b" \
    --env NGUI_PROVIDER_LLAMA_URL="http://host.containers.internal:5001" \
    quay.io/next-gen-ui/mcp

Configuration Using Environment File

Create a .env file:

# .env file
MCP_PORT=5000
MCP_HOST=0.0.0.0
MCP_TRANSPORT=streamable-http
MCP_STRUCTURED_OUTPUT_ENABLED="false"
NGUI_COMPONENT_SYSTEM=json
NGUI_PROVIDER=langchain
NGUI_MODEL=gpt-4o
NGUI_PROVIDER_API_KEY=your-api-key-here

Run with environment file:

podman run --rm -it -p 5000:5000 --env-file .env quay.io/next-gen-ui/mcp

Network Configuration

For local development connecting to services running on the host machine:

Use host.containers.internal to access host services (works with Podman and Docker Desktop)
For Linux with Podman, you may need to use host.docker.internal or the host's IP address
Ensure the target services (like Ollama) are accessible from containers