Configuration

The Next Gen UI Agent can be configured in two ways: programmatically using Python dictionaries or declaratively using YAML configuration files.

Configuration Options

inference [InferenceBase, required]

The LLM inference engine to use for component selection and configuration.

The inference parameter accepts any class that extends InferenceBase. The Next Gen UI Agent provides the following built-in inference implementations:

• next_gen_ui_agent.model.LangChainModelInference: Wrapper for LangChain-compatible language models. Supports any LangChain LLM implementation including ChatOllama, ChatOpenAI, ChatAnthropic, and other LangChain model providers.
• next_gen_ui_llama_stack.LlamaStackAgentInference: Integration with Standalone Llama Stack
• next_gen_ui_llama_stack_embedded.LlamaStackEmbeddedAsyncAgentInference: Integration with Embedded Llama Stack

component_system [str, optional]

Component system for rendering (default: "json")

unsupported_components [bool, optional]

Whether to allow unsupported UI components (default: False)

component_selection_strategy [str, optional]

Strategy for component selection

• one_llm_call / default: Uses single LLM call for component selection and configuration
• two_llm_calls: Uses two-step LLM process - first selects component type, then configures it

hand_build_components_mapping [dict[str, str], optional]

Mapping from input data types to hand-built component types

See Hand Build Components chapter

Programmatic Configuration

Usage with Inference Configuration

from next_gen_ui_agent import NextGenUIAgent, LangChainModelInference
from langchain_ollama import ChatOllama

# Configure LLM inference
llm = ChatOllama(model="llama3.2")
inference = LangChainModelInference(llm)

# Create configuration with inference
config = {
    "inference": inference,
    "component_system": "json",
    "component_selection_strategy": "default",
    "unsupported_components": True
}

agent = NextGenUIAgent(config)

With Hand-Built Components

config = {
    "component_system": "json",
    "hand_build_components_mapping": {
        "movies.movie-detail": "movies:movie-detail-view",
        "movies.movies-list": "movies:movies-list-view",
    }
}

agent = NextGenUIAgent(config)

YAML Configuration

Basic YAML Configuration

Create a YAML configuration file:

---
component_system: json
unsupported_components: false
component_selection_strategy: default

hand_build_components_mapping:
  movies.movie-detail: movies:movie-detail-view
  movies.movies-list: movies:movies-list-view,

Loading YAML Configuration

From File Path

from next_gen_ui_agent import NextGenUIAgent
from next_gen_ui_agent.agent_config import read_config_yaml_file

# Load configuration from file
config = read_config_yaml_file("path/to/config.yaml")
agent = NextGenUIAgent(config)

From YAML String

from next_gen_ui_agent import NextGenUIAgent

yaml_config = """
component_system: json
component_selection_strategy: two_llm_calls
unsupported_components: true

hand_build_components_mapping:
  namespace.list: namespace-table
  pod.details: pod-card
"""

# Pass YAML string directly to constructor
agent = NextGenUIAgent(yaml_config)