Agents, Sessions & Workflows API¶

API reference for litestar_mcp.agent: declarative Agent and AgentGroup specs, AgentRuntime and TurnBudget, AgentChatController, pluggable SessionStore implementations (MemorySessionStore, LitestarStoreSessionStore), model adapters (ModelClient, GoogleGenAIClient, MockModelClient), SSE AgentStreamFrame streaming, and in-process DynamicWorkflow / WorkflowEngine DAG execution.

For guides and patterns, see Agent Runtime & Turn Budgets, Agent Groups & Handoffs, Sessions & SSE Streaming, and Durable DAG Workflows.

Agent & Group Specifications¶

Agent ¶

Agent definition with instructions, tools, skills, and an optional model binding.

Parameters:

Name	Type	Description	Default
`name`	`str`	Unique identifier of the agent within an application or `AgentGroup`.	required
`description`	`str`	Human-readable summary of the agent's purpose and capabilities.	`''`
`instructions`	`str`	Base system instructions guiding the agent's behavior.	`''`
`tools`	`Sequence[Tool \| Callable[..., Any]]`	Tools or plain callables exposed directly to the agent.	`()`
`skills`	`Sequence[SkillController \| type[SkillController]]`	`SkillController` classes or instances contributing tools and instructions.	`()`
`model`	`ModelClient \| str \| None`	Bound `ModelClient` instance or model identifier string.	`None`

Attributes:

Name	Type	Description
`name`	`str`	Unique identifier of the agent within an application or `AgentGroup`.
`description`	`str`	Human-readable summary of the agent's purpose and capabilities.
`instructions`	`str`	Base system instructions guiding the agent's behavior.
`tools`	`Sequence[Tool \| Callable[..., Any]]`	Tools or plain callables exposed directly to the agent.
`skills`	`Sequence[SkillController \| type[SkillController]]`	`SkillController` classes or instances contributing tools and instructions.
`model`	`ModelClient \| str \| None`	Bound `ModelClient` instance or model identifier string.
`tool_set`	`tuple[Tool, ...]`	Return the agent's own tools followed by skill tools.
`skill_instances`	`tuple[SkillController, ...]`	Return the instantiated skill controllers attached to this agent.
`combined_instructions`	`str`	Return agent instructions joined with skill instructions by blank lines.

AgentGroup ¶

Coordinator and specialist agents with a group-owned transfer_to_agent tool.

Attributes:

Name	Type	Description
`specialists`	`dict[str, Agent]`	Default: `{s.name: s for s in specialists}`

tools_for ¶

tools_for(agent)

Return tools available to an agent in this group.

get_agent ¶

get_agent(name)

Retrieve an agent by name from coordinator or specialist roster.

resolve_transfer ¶

resolve_transfer(call)

Resolve a tool call into an agent transfer target if applicable.

AgentMessage ¶

Bases: Struct

Provider-neutral conversation message.

Parameters:

Name	Type	Description	Default
`role`	`Literal['system', 'user', 'assistant', 'tool']`	Message author role (`system`, `user`, `assistant`, or `tool`).	required
`content`	`str`	Primary text content of the message.	`''`
`tool_calls`	`list[ToolCall]`	Built-in mutable sequence. If no argument is given, the constructor creates a new empty list. The argument must be an iterable if specified.	`<dynamic>`
`tool_results`	`list[ToolResult]`	Built-in mutable sequence. If no argument is given, the constructor creates a new empty list. The argument must be an iterable if specified.	`<dynamic>`
`agent_name`	`str \| None`	Optional name of the agent that produced or handled this message.	`None`

Attributes:

Name	Type	Description
`role`	`Literal['system', 'user', 'assistant', 'tool']`	Message author role (`system`, `user`, `assistant`, or `tool`).
`content`	`str`	Primary text content of the message.
`tool_calls`	`list[ToolCall]`	Tool invocations requested by the assistant in this message.
`tool_results`	`list[ToolResult]`	Tool execution results returned in a `tool` role message.
`agent_name`	`str \| None`	Optional name of the agent that produced or handled this message.

Agent Runtime & Turn Budgets¶

AgentRuntime ¶

Bases: InitPluginProtocol

Runs agent turns with budgets, owner-scoped sessions, and streamed frames.

Register the runtime in Litestar(plugins=[...]) to provide it as the runtime dependency used by AgentChatController.

on_app_init ¶

on_app_init(app_config)

Provide this runtime as the application's runtime dependency.

Returns:

Type	Description
`AppConfig`	The updated application configuration.

get_history ¶

get_history(context, session_id)

Return conversation history scoped to connection owner.

produce ¶

produce(turn, send)

Execute turn loop as a producer coroutine emitting frames to the send stream.

open_stream ¶

open_stream(turn)

Open a background task group streaming frames to an in-memory receive stream.

stream ¶

stream(turn)

Yield stream frames sequentially from open_stream.

run_turn ¶

run_turn(turn)

Run turn to completion and return TurnResponse.

TurnRequest ¶

Input parameters for one agent turn execution.

Parameters:

Name	Type	Description	Default
`user_message`	`str`	User prompt text initiating the agent turn.	`''`
`session_id`	`str \| None`	Optional session identifier to resume an existing conversation.	`None`
`turn_id`	`str \| None`	Optional caller-supplied identifier for the turn.	`None`
`dynamic_context`	`str \| None`	Optional per-turn context prepended to the active agent's instructions.	`None`
`context`	`ToolContext \| None`	Request-scoped tool execution and principal ownership context.	`None`

Attributes:

Name	Type	Description
`user_message`	`str`	User prompt text initiating the agent turn.
`session_id`	`str \| None`	Optional session identifier to resume an existing conversation.
`turn_id`	`str \| None`	Optional caller-supplied identifier for the turn.
`dynamic_context`	`str \| None`	Optional per-turn context prepended to the active agent's instructions.
`context`	`ToolContext \| None`	Request-scoped tool execution and principal ownership context.

TurnResponse ¶

Aggregated result of a completed agent turn.

Parameters:

Name	Type	Description	Default
`session_id`	`str`	Session identifier under which the turn history was persisted.	required
`turn_id`	`str`	Unique identifier for the completed turn.	required
`output`	`str`	Concatenated assistant text output produced across the turn.	required
`agent_name`	`str`	Name of the active agent at the end of the turn.	required
`messages`	`list[AgentMessage]`	Built-in mutable sequence. If no argument is given, the constructor creates a new empty list. The argument must be an iterable if specified.	`<dynamic>`
`token_usage`	`dict[str, int]`	dict() -> new empty dictionary dict(mapping) -> new dictionary initialized from a mapping object's (key, value) pairs dict(iterable) -> new dictionary initialized as if via: d = {} for k, v in iterable: d[k] = v dict(**kwargs) -> new dictionary initialized with the name=value pairs in the keyword argument list. For example: dict(one=1, two=2)	`<class 'dict'>`

Attributes:

Name	Type	Description
`session_id`	`str`	Session identifier under which the turn history was persisted.
`turn_id`	`str`	Unique identifier for the completed turn.
`output`	`str`	Concatenated assistant text output produced across the turn.
`agent_name`	`str`	Name of the active agent at the end of the turn.
`messages`	`list[AgentMessage]`	Full conversation history including the newly appended turn messages.
`token_usage`	`dict[str, int]`	Aggregated token usage counts for prompt, completion, and thought tokens.
`content`	`str`	Alias for output text produced by the turn.

TurnBudget ¶

Configurable execution limits for a single agent turn.

Parameters:

Name	Type	Description	Default
`max_turns`	`int`	Maximum number of model-invocation iterations allowed in a single turn.	`15`
`max_tool_calls`	`int`	Maximum total tool calls allowed across all iterations of a single turn.	`50`
`timeout_seconds`	`float`	Wall-clock timeout in seconds for the entire turn execution.	`300.0`

Attributes:

Name	Type	Description
`max_turns`	`int`	Maximum number of model-invocation iterations allowed in a single turn.
`max_tool_calls`	`int`	Maximum total tool calls allowed across all iterations of a single turn.
`timeout_seconds`	`float`	Wall-clock timeout in seconds for the entire turn execution.

BudgetExceededError ¶

Bases: LitestarMCPError

Raised when an agent turn exceeds its configured turn, tool-call, or timeout budget.

HTTP & SSE Chat Controller¶

AgentChatController ¶

Bases: Controller

SSE chat, synchronous turns, and owner-scoped history over an injected AgentRuntime.

Attributes:

Name	Type	Description
`ping_interval`	`float`	Default: `15.0`

chat ¶

chat(request, data, runtime)

Stream an agent turn over Server-Sent Events from a typed chat payload.

create_turn ¶

create_turn(request, data, runtime)

Execute an agent turn synchronously returning final TurnResult.

get_session ¶

get_session(request, session_id, runtime)

Return conversation history for a session scoped to caller identity.

ChatRequest ¶

Bases: Struct

Body payload for agent chat streaming and turn creation.

Parameters:

Name	Type	Description	Default
`message`	`str`	User prompt text for the agent turn.	required
`session_id`	`str \| None`	Optional existing session identifier to continue a conversation.	`None`
`dynamic_context`	`str \| None`	Optional per-turn context prepended to the agent's system instructions.	`None`

Attributes:

Name	Type	Description
`message`	`str`	User prompt text for the agent turn.
`session_id`	`str \| None`	Optional existing session identifier to continue a conversation.
`dynamic_context`	`str \| None`	Optional per-turn context prepended to the agent's system instructions.

TurnResult ¶

Bases: Struct

Aggregated turn result returned by POST /agent/turns.

Parameters:

Name	Type	Description	Default
`session_id`	`str`	Identifier of the conversation session for this turn.	required
`turn_id`	`str`	Unique identifier assigned to this turn execution.	required
`agent_name`	`str`	Name of the agent that produced the final turn output.	required
`output`	`str`	Concatenated assistant text output produced during the turn.	required
`messages`	`list[AgentMessage]`	Updated conversation message history after the turn completes.	required
`token_usage`	`dict[str, int]`	Accumulated token usage counts keyed by token category.	required

Attributes:

Name	Type	Description
`session_id`	`str`	Identifier of the conversation session for this turn.
`turn_id`	`str`	Unique identifier assigned to this turn execution.
`agent_name`	`str`	Name of the agent that produced the final turn output.
`output`	`str`	Concatenated assistant text output produced during the turn.
`messages`	`list[AgentMessage]`	Updated conversation message history after the turn completes.
`token_usage`	`dict[str, int]`	Accumulated token usage counts keyed by token category.

SessionHistory ¶

Bases: Struct

Conversation history payload returned by GET /agent/sessions/{session_id}.

Parameters:

Name	Type	Description	Default
`session_id`	`str`	Identifier of the retrieved conversation session.	required
`messages`	`list[AgentMessage]`	Ordered list of messages stored in the session.	required

Attributes:

Name	Type	Description
`session_id`	`str`	Identifier of the retrieved conversation session.
`messages`	`list[AgentMessage]`	Ordered list of messages stored in the session.

SessionNotFoundError ¶

Bases: NotFoundException

Exception raised when a requested session is not found.

Session Stores¶

SessionStore ¶

Bases: Protocol

Persist conversation history keyed by owner and session id.

load ¶

load(owner, session_id)

Load messages for an owner and session id, returning None if nonexistent.

save ¶

save(owner, session_id, messages)

Persist messages for an owner and session id.

lock ¶

lock(owner, session_id)

Acquire a per-session lock context manager.

MemorySessionStore ¶

Process-local LRU session store with TTL expiry and per-session locks.

lock ¶

lock(owner, session_id)

Acquire per-session lock.

load ¶

load(owner, session_id)

Load messages if entry exists and has not expired.

save ¶

save(owner, session_id, messages)

Store messages and evict oldest when capacity exceeded.

LitestarStoreSessionStore ¶

Session store over a Litestar Store backend.

lock ¶

lock(owner, session_id)

Acquire per-session lock.

load ¶

load(owner, session_id)

Load and decode conversation messages from storage backend.

save ¶

save(owner, session_id, messages)

Encode and persist conversation messages to storage backend.

Model Clients & Streaming Frames¶

ModelClient ¶

Bases: Protocol

Asynchronous protocol implemented by all LLM model adapters.

stream_turn ¶

stream_turn(*, system_instruction, messages, tools)

Stream an agent turn yielding text, thoughts, tool calls, and usage deltas.

GoogleGenAIClient ¶

First-party Google GenAI model client wrapping google.genai with httpx2 support.

Attributes:

Name	Type	Description
`provider_name`	`str`	Return canonical OpenTelemetry provider name.

stream_turn ¶

stream_turn(*, system_instruction, messages, tools)

Stream an agent turn using Google GenAI SDK.

MockModelClient ¶

Deterministic scriptable model client for fast offline unit tests.

Attributes:

Name	Type	Description
`responses`	`list[list[ModelDelta]]`	Default: `responses or []`
`call_history`	`list[dict[str, Any]]`	Default: `[]`

stream_turn ¶

stream_turn(*, system_instruction, messages, tools)

Record turn inputs and yield queued scripted responses.

ModelDelta ¶

Incremental chunk emitted by a ModelClient during an agent turn.

Parameters:

Name	Type	Description	Default
`event_type`	`ModelEventType`	Kind of incremental model event (`delta`, `thought`, `tool_call`, or `usage`).	required
`text`	`str \| None`	Incremental assistant text or thought content.	`None`
`thought`	`str \| None`	Reasoning or thinking trace text when emitted by a thinking model.	`None`
`tool_name`	`str \| None`	Name of the tool requested by the model for a `tool_call` event.	`None`
`call_id`	`str \| None`	Unique identifier correlating a tool call with its result.	`None`
`arguments`	`dict[str, Any] \| None`	Decoded JSON arguments mapping passed to the requested tool.	`None`
`prompt_tokens`	`int \| None`	Input prompt token count reported in a `usage` event.	`None`
`completion_tokens`	`int \| None`	Output completion token count reported in a `usage` event.	`None`
`provider_metadata`	`dict[str, dict[str, Any]]`	dict() -> new empty dictionary dict(mapping) -> new dictionary initialized from a mapping object's (key, value) pairs dict(iterable) -> new dictionary initialized as if via: d = {} for k, v in iterable: d[k] = v dict(**kwargs) -> new dictionary initialized with the name=value pairs in the keyword argument list. For example: dict(one=1, two=2)	`<class 'dict'>`
`thought_tokens`	`int \| None`	Reasoning token count reported in a `usage` event.	`None`

Attributes:

Name	Type	Description
`event_type`	`ModelEventType`	Kind of incremental model event (`delta`, `thought`, `tool_call`, or `usage`).
`text`	`str \| None`	Incremental assistant text or thought content.
`thought`	`str \| None`	Reasoning or thinking trace text when emitted by a thinking model.
`tool_name`	`str \| None`	Name of the tool requested by the model for a `tool_call` event.
`call_id`	`str \| None`	Unique identifier correlating a tool call with its result.
`arguments`	`dict[str, Any] \| None`	Decoded JSON arguments mapping passed to the requested tool.
`prompt_tokens`	`int \| None`	Input prompt token count reported in a `usage` event.
`completion_tokens`	`int \| None`	Output completion token count reported in a `usage` event.
`provider_metadata`	`dict[str, dict[str, Any]]`	Provider-specific metadata such as call identifiers and thought signatures.
`thought_tokens`	`int \| None`	Reasoning token count reported in a `usage` event.

AgentStreamFrame ¶

Bases: Struct

Typed SSE frame emitted during agent execution.

Parameters:

Name	Type	Description	Default
`event_type`	`Literal['session', 'thought', 'delta', 'tool_call', 'tool_result', 'agent_transfer', 'ping', 'complete', 'error']`	Frame event discriminator (`session`, `thought`, `delta`, `tool_call`, `tool_result`, `agent_transfer`, `ping`, `complete`, or `error`).	required
`turn_id`	`str`	Unique identifier of the turn that emitted this frame.	required
`seq`	`int`	Monotonically increasing 1-based sequence number within the turn stream.	`0`
`agent_name`	`str \| None`	Name of the active agent associated with this frame.	`None`
`text`	`str \| None`	Incremental text chunk for `delta` or `thought` frames.	`None`
`tool_name`	`str \| None`	Tool name for `tool_call` and `tool_result` frames.	`None`
`call_id`	`str \| None`	Tool call correlation identifier for `tool_call` and `tool_result` frames.	`None`
`payload`	`dict[str, Any] \| None`	Structured event payload for session, tool, transfer, completion, or error frames.	`None`

Attributes:

Name	Type	Description
`event_type`	`StreamEventType`	Frame event discriminator (`session`, `thought`, `delta`, `tool_call`, `tool_result`, `agent_transfer`, `ping`, `complete`, or `error`).
`turn_id`	`str`	Unique identifier of the turn that emitted this frame.
`seq`	`int`	Monotonically increasing 1-based sequence number within the turn stream.
`agent_name`	`str \| None`	Name of the active agent associated with this frame.
`text`	`str \| None`	Incremental text chunk for `delta` or `thought` frames.
`tool_name`	`str \| None`	Tool name for `tool_call` and `tool_result` frames.
`call_id`	`str \| None`	Tool call correlation identifier for `tool_call` and `tool_result` frames.
`payload`	`dict[str, Any] \| None`	Structured event payload for session, tool, transfer, completion, or error frames.

to_sse_message ¶

to_sse_message(type_encoders=None)

Encode as an SSE message whose id is seq when non-zero and whose event is event_type.

from_sse ¶

from_sse(event)

Decode an SSE event, taking seq from the event id when the payload omits it.

iter_agent_stream ¶

iter_agent_stream(response, *, max_event_size=1048576)

Consume a streaming httpx2.Response using httpx2.EventSource and yield AgentStreamFrame objects.

DAG Workflows (`DynamicWorkflow` & `WorkflowEngine`)¶

DynamicWorkflow ¶

DAG-based workflow connecting agents and discrete task handlers.

Attributes:

Name	Type	Description
`nodes`	`dict[str, WorkflowNode]`	Default: `{}`

add_node ¶

add_node(
    name_or_node,
    *,
    agent=None,
    handler=None,
    depends_on=None,
    dependencies=None,
    prompt=None,
    timeout_seconds=300.0,
    condition=None,
)

Add a step to the workflow with explicit dependencies.

get_ready_nodes ¶

get_ready_nodes(completed)

Return names of all nodes whose dependencies are satisfied and not yet completed.

WorkflowNode ¶

A single execution node within a DynamicWorkflow DAG.

Parameters:

Name	Type	Description	Default
`name`	`str`	Unique node name within the workflow DAG.	required
`agent`	`Agent \| None`	Optional `Agent` executed when this node runs.	`None`
`handler`	`Callable[..., Any] \| Agent \| None`	Optional synchronous or asynchronous callable accepting `WorkflowContext`.	`None`
`depends_on`	`tuple[str, ...] \| Sequence[str]`	Names of upstream nodes that must complete before this node runs.	`()`
`prompt`	`Callable[list, str] \| None`	Optional callable building the prompt string for an `agent` node from `WorkflowContext`.	`None`
`timeout_seconds`	`float`	Maximum execution time in seconds allowed for this node.	`300.0`
`condition`	`Callable[list, bool] \| None`	Optional predicate evaluated against `WorkflowContext`; when false, the node is skipped.	`None`

Attributes:

Name	Type	Description
`name`	`str`	Unique node name within the workflow DAG.
`agent`	`Agent \| None`	Optional `Agent` executed when this node runs.
`handler`	`Callable[..., Any] \| Agent \| None`	Optional synchronous or asynchronous callable accepting `WorkflowContext`.
`depends_on`	`tuple[str, ...] \| Sequence[str]`	Names of upstream nodes that must complete before this node runs.
`prompt`	`Callable[[WorkflowContext], str] \| None`	Optional callable building the prompt string for an `agent` node from `WorkflowContext`.
`timeout_seconds`	`float`	Maximum execution time in seconds allowed for this node.
`condition`	`Callable[[WorkflowContext], bool] \| None`	Optional predicate evaluated against `WorkflowContext`; when false, the node is skipped.
`dependencies`	`list[str]`	Backward-compatible list representation of node dependencies.

WorkflowEngine ¶

Concurrent in-process DAG execution engine for DynamicWorkflow pipelines.

execute_node ¶

execute_node(workflow, node_name, context_or_state)

Execute a single workflow node against current workflow context or state.

validate_dependencies ¶

validate_dependencies(workflow)

Validate that all node dependencies exist and graph is acyclic.

execute_run ¶

execute_run(workflow, initial_state=None)

Execute workflow in DAG topological order via ready queue scheduling.

run ¶

run(workflow, initial_state=None)

Execute the workflow to completion and return the final state dictionary.

WorkflowContext ¶

Execution context tracking state and step outputs across a workflow run.

Parameters:

Name	Type	Description	Default
`run_id`	`str`	Unique identifier for the workflow execution run.	`'32ea9688df6f47c9b049732194fc8c1b'`
`state`	`dict[str, Any]`	dict() -> new empty dictionary dict(mapping) -> new dictionary initialized from a mapping object's (key, value) pairs dict(iterable) -> new dictionary initialized as if via: d = {} for k, v in iterable: d[k] = v dict(**kwargs) -> new dictionary initialized with the name=value pairs in the keyword argument list. For example: dict(one=1, two=2)	`<class 'dict'>`
`node_outputs`	`dict[str, Any]`	dict() -> new empty dictionary dict(mapping) -> new dictionary initialized from a mapping object's (key, value) pairs dict(iterable) -> new dictionary initialized as if via: d = {} for k, v in iterable: d[k] = v dict(**kwargs) -> new dictionary initialized with the name=value pairs in the keyword argument list. For example: dict(one=1, two=2)	`<class 'dict'>`
`step_results`	`dict[str, StepResult]`	dict() -> new empty dictionary dict(mapping) -> new dictionary initialized from a mapping object's (key, value) pairs dict(iterable) -> new dictionary initialized as if via: d = {} for k, v in iterable: d[k] = v dict(**kwargs) -> new dictionary initialized with the name=value pairs in the keyword argument list. For example: dict(one=1, two=2)	`<class 'dict'>`

Attributes:

Name	Type	Description
`run_id`	`str`	Unique identifier for the workflow execution run.
`state`	`dict[str, Any]`	Mutable shared state dictionary accessible to all workflow nodes.
`node_outputs`	`dict[str, Any]`	Mapping of completed node names to their raw output values.
`step_results`	`dict[str, StepResult]`	Mapping of executed node names to their `StepResult` records.

get ¶

get(key, default=None)

Retrieve a value from workflow state or node outputs.

StepResult ¶

Result record for a single executed workflow step.

Parameters:

Name	Type	Description	Default
`node_name`	`str`	Name of the workflow node that produced this result.	required
`output`	`Any`	Value returned by the node handler or agent turn.	required
`status`	`Literal['completed', 'skipped', 'failed']`	Terminal execution status of the step (`completed`, `skipped`, or `failed`).	`'completed'`
`error`	`str \| None`	Optional error message when `status` is `failed`.	`None`

Attributes:

Name	Type	Description
`node_name`	`str`	Name of the workflow node that produced this result.
`output`	`Any`	Value returned by the node handler or agent turn.
`status`	`StepStatus`	Terminal execution status of the step (`completed`, `skipped`, or `failed`).
`error`	`str \| None`	Optional error message when `status` is `failed`.

Agents, Sessions & Workflows API¶

Agent & Group Specifications¶

Agent ¶

AgentGroup ¶

tools_for ¶

get_agent ¶

resolve_transfer ¶

AgentMessage ¶

Agent Runtime & Turn Budgets¶

AgentRuntime ¶

on_app_init ¶

get_history ¶

produce ¶

open_stream ¶

stream ¶

run_turn ¶

TurnRequest ¶

TurnResponse ¶

TurnBudget ¶

BudgetExceededError ¶

HTTP & SSE Chat Controller¶

AgentChatController ¶

chat ¶

create_turn ¶

get_session ¶

ChatRequest ¶

TurnResult ¶

SessionHistory ¶

SessionNotFoundError ¶

Session Stores¶

SessionStore ¶

load ¶

save ¶

lock ¶

MemorySessionStore ¶

lock ¶

load ¶

save ¶

LitestarStoreSessionStore ¶

lock ¶

load ¶

save ¶

Model Clients & Streaming Frames¶

ModelClient ¶

stream_turn ¶

GoogleGenAIClient ¶

stream_turn ¶

MockModelClient ¶

stream_turn ¶

ModelDelta ¶

AgentStreamFrame ¶

to_sse_message ¶

from_sse ¶

iter_agent_stream ¶

DAG Workflows (DynamicWorkflow & WorkflowEngine)¶

DynamicWorkflow ¶

add_node ¶

get_ready_nodes ¶

WorkflowNode ¶

WorkflowEngine ¶

execute_node ¶

validate_dependencies ¶

execute_run ¶

run ¶

WorkflowContext ¶

get ¶

StepResult ¶

DAG Workflows (`DynamicWorkflow` & `WorkflowEngine`)¶