llmcat

High-level tools to create interactive command-line REPLs for LLMs and LangGraph

Example

DEFAULT_CONFIG = {
    # Chat
    'max_width': 100,

    # Chat Session
    'temperature': 1.2,
    'model_option_name': 'gemma-3-1b',
    'output_mode': 'simple',

    # Hugging Face LLM
    'device': 'cpu',
    'return_full_text': False,
}

model_options = [
    HuggingFaceModelOption(name='gemma-3-1b', model_id="google/gemma-3-1b-it"),
    OpenAIModelOption(name='gpt-5-nano', model='gpt-5-nano-2025-08-07'),
]
commands = [
    ConfigsCommand(),
    ModelsCommand(),
    ResetConfigCommand(),
]
cs = ChatSession(
    available_models=model_options,
    default_config=DEFAULT_CONFIG,
    config_path=CONFIG_PATH,
)
chat_loop = ChatPromptLoop(cs, commands)
chat_loop.initialize()
chat_loop.run()