Task Tools¶

The Task Tools component provides LangChain tools for creating, listing, updating, and cancelling tasks within epics. Tasks are the individual units of work that agents execute, each trackable with status, dependencies, and budget.

Component type: task_tools

How It Works¶

This component registers four tools with the parent agent. Tasks belong to epics and support dependency chains -- a task can depend on other tasks within the same epic, and will be automatically set to blocked status until its dependencies complete. When a task is marked completed, any downstream tasks that depended on it are automatically unblocked.

Ports¶

Outputs¶

Configuration¶

This tool has no configurable settings.

Usage¶

Connect this tool to an agent via the green diamond tool handle. It is typically paired with Epic Tools and Spawn & Await:

flowchart LR
    Epic[Epic Tools] -.->|tool| Agent
    Task[Task Tools] -.->|tool| Agent
    SA[Spawn & Await] -.->|tool| Agent
    Model[AI Model] -.->|llm| Agent
    Trigger[Chat Trigger] --> Agent

Tools Provided¶

This component registers four tools with the agent:

create_task¶

Create a new task within an epic.

create_task(
    epic_id: str,
    title: str,
    description: str = "",
    tags: str = "",
    depends_on: str = "",
    priority: int = 2,
    estimated_tokens: int | None = None,
    max_retries: int = 2,
) -> str

Returns: {"success": true, "task_id": "tsk-abc123", "status": "pending"}

list_tasks¶

List tasks within an epic, optionally filtered by status or tags.

list_tasks(
    epic_id: str,
    status: str | None = None,
    tags: str | None = None,
    limit: int = 20,
) -> str

Returns: JSON with tasks list (each with id, title, status, priority, depends_on, execution_id) and total count.

update_task¶

Update a task's fields. Only provided fields are changed.

update_task(
    task_id: str,
    status: str | None = None,
    title: str | None = None,
    description: str | None = None,
    priority: int | None = None,
    result_summary: str | None = None,
    error_message: str | None = None,
    notes: str | None = None,
) -> str

Returns: {"success": true, "task_id": "tsk-abc123", "status": "completed"}

cancel_task¶

Cancel a task and optionally its linked execution.

cancel_task(task_id: str, reason: str = "") -> str

Returns: {"success": true, "task_id": "tsk-abc123", "execution_cancelled": true}

If the task has a linked workflow execution that is still pending or running, it is cancelled as well.

Example¶

An orchestrator agent managing a multi-step project:

Tool call: create_task(
    epic_id="ep-w3ath3r",
    title="Create weather API integration workflow",
    description="Build a workflow that calls the OpenWeatherMap API",
    tags="api,integration",
    priority=1
)
Result: {"success": true, "task_id": "tsk-api001", "status": "pending"}

Tool call: create_task(
    epic_id="ep-w3ath3r",
    title="Create data storage workflow",
    description="Build a workflow that stores weather data",
    depends_on="tsk-api001",
    priority=2
)
Result: {"success": true, "task_id": "tsk-store01", "status": "blocked"}

The second task is automatically blocked because it depends on the first. When the agent marks tsk-api001 as completed, tsk-store01 is automatically unblocked and moves to pending.