Class TaskConfig

langroid/agent/task.py:110–172

Configuration for a Task. This is a container for any params that we didn't include in the task `__init__` method. We may eventually move all the task __init__ params to this class, analogous to how we have config classes for `Agent`, `ChatAgent`, `LanguageModel`, etc. Attributes:

class TaskConfig(BaseModel):
    """Configuration for a Task. This is a container for any params that
    we didn't include in the task `__init__` method.
    We may eventually move all the task __init__ params to this class, analogous to how
    we have config classes for `Agent`, `ChatAgent`, `LanguageModel`, etc.

    Attributes:
        inf_loop_cycle_len (int): max exact-loop cycle length: 0 => no inf loop test
        inf_loop_dominance_factor (float): dominance factor for exact-loop detection
        inf_loop_wait_factor (int): wait this * cycle_len msgs before loop-check
        restart_as_subtask (bool): whether to restart *every* run of this task
            when run as a subtask.
        addressing_prefix (str): "@"-like prefix an agent can use to address other
            agents, or entities of the agent. E.g., if this is "@", the addressing
            string would be "@Alice", or "@user", "@llm", "@agent", etc.
            If this is an empty string, then addressing is disabled.
            Default is empty string "".
            CAUTION: this is a deprecated practice, since normal prompts
            can accidentally contain such addressing prefixes, and will break
            your runs. This could happen especially when your prompt/context
            contains code, but of course could occur in normal text as well.
            Instead, use the `RecipientTool` to have agents address other agents or
            entities. If you do choose to use `addressing_prefix`, the recommended
            setting is to use `langroid.utils.constants.AT`, which currently is "|@|".
            Note that this setting does NOT affect the use of `constants.SEND_TO` --
            this is always enabled since this is a critical way for responders to
            indicate that the message should be sent to a specific entity/agent.
            (Search for "SEND_TO" in the examples/ dir to see how this is used.)
        allow_subtask_multi_oai_tools (bool): whether to allow multiple OpenAI
            tool-calls to be sent to a sub-task.
        recognize_string_signals (bool): whether to recognize string-based signaling
            like DONE, SEND_TO, PASS, etc. Default is True, but note that we don't need
            to use string-based signaling, and it is recommended to use the
            new Orchestration tools instead (see agent/tools/orchestration.py),
            e.g. DoneTool, SendTool, etc.
            Note: this is distinct from
            ``ChatAgentConfig.recognize_recipient_in_content``, which controls
            whether LLM response text is parsed for ``TO[<recipient>]:`` and
            JSON ``{"recipient": ...}`` patterns at the Agent level.
            To fully disable all text-based routing, set both to False.
        done_if_tool (bool): whether to consider the task done if the pending message
            contains a Tool attempt by the LLM
            (including tools not handled by the agent).
            Default is False.
        done_sequences (List[DoneSequence]): List of event sequences that trigger task
            completion. Task is done if ANY sequence matches the recent event history.
            Each sequence is checked against the message parent chain.
            Tool classes can be referenced in sequences like "T[MyToolClass]".

    """

    inf_loop_cycle_len: int = 10
    inf_loop_dominance_factor: float = 1.5
    inf_loop_wait_factor: int = 5
    restart_as_subtask: bool = False
    logs_dir: str = "logs"
    enable_loggers: bool = True
    enable_html_logging: bool = True