DasshhRuntime

This page lays out all the methods and attributes available on the DasshhRuntime class.

Tip

DasshhRuntime is the core engine that handles AI model interactions, tool execution, and session management in Dasshh.

attr model_config

The model configuration dictionary containing all LiteLLM parameters

model_config: dict = {}

This dictionary contains the complete model configuration (via litellm_params key) including:

• model: The model provider and name (e.g., "gemini/gemini-2.0-flash")
• api_key: The API key for the chosen provider
• base_url: The base URL for the API (optional)
• api_version: The API version to use (optional)
• temperature: The temperature for the model (default: 1.0)
• top_p: The top_p value for the model (default: 1.0)
• max_tokens: The maximum number of tokens to generate
• max_completion_tokens: The maximum number of tokens for completion

attr system_prompt

The system prompt for the runtime

system_prompt: str = "Your name is Dasshh..."

attr skip_summarization

Whether to skip summarization after a tool call

skip_summarization: bool = False

property get_system_prompt

get_system_prompt() -> dict

Returns the system prompt as a formatted message dictionary

Returns:

Type	Default	Description
dict		System prompt formatted as a message with role and content

method init

__init__(
    *,
    session_service: SessionService,
    model_config: dict = {},
    system_prompt: str = "",
    skip_summarization: bool = False,
)

Initialize the DasshhRuntime with configuration parameters

Parameters:

Param	Default	Description
session_service		The SessionService instance for managing conversations and events
model_config	{}	Dictionary containing LiteLLM model configuration parameters
system_prompt	""	Custom system prompt for the assistant
skip_summarization	False	Whether to skip summarization after tool calls

method start

async start()

Start the runtime worker that processes the query queue

method stop

async stop()

Stop the runtime worker and cancel any pending operations

method submit_query

async submit_query(
    *,
    message: str,
    session_id: str,
    post_message_callback: Callable,
) -> None

Submit a query to the runtime for processing

Parameters:

Param	Default	Description
message		The user message to send to the AI
session_id		The session ID to associate with this conversation
post_message_callback		Callback function to send UI events back to the interface

Returns:

Type	Default	Description
None		This method is async and doesn't return a value

method _generate_prompt

_generate_prompt(context: InvocationContext) -> List[dict]

Generate the complete prompt including system message and conversation history

Parameters:

Param	Default	Description
context		The invocation context containing session and message information

Returns:

Type	Default	Description
List[dict]		List of message dictionaries formatted for the AI model

method _process_queue

async _process_queue()

Main worker loop that processes queries from the queue

method _run_async

async _run_async(context: InvocationContext) -> AsyncGenerator[ModelResponse, None]

Execute the AI completion request and yield streaming responses

Parameters:

Param	Default	Description
context		The invocation context for this query

Returns:

Type	Default	Description
AsyncGenerator		Stream of model responses

method _handle_tool_calls

async _handle_tool_calls(
    context: InvocationContext,
    tool_calls: list[ChatCompletionDeltaToolCall],
) -> None

Handle tool calls from the AI model

Parameters:

Param	Default	Description
context		The invocation context for this query
tool_calls		List of tool calls from the AI model

Configuration Integration

The DasshhRuntime receives its model configuration from the Dasshh app, which manages multiple model configurations:

# Example of how the runtime is configured
runtime = DasshhRuntime(
    session_service=session_service,
    model_config={
        "model_name": "gemini-flash",
        "litellm_params": {
            "model": "gemini/gemini-2.0-flash",
            "api_key": "your-api-key",
            "temperature": 0.7,
            "max_tokens": 2000
        }
    },
    system_prompt="Your custom system prompt",
    skip_summarization=False
)

The model configuration corresponds to the litellm_params section of a model entry in the configuration file.