iflytek/astron-agent · error · CustomException

WORKFLOW_EXECUTION_ERROR

WORKFLOW_EXECUTION_ERROR

Error message

Flow output mode not configured for flow_id: {self.flowId}

What it means

WORKFLOW_EXECUTION_ERROR from flow_node.req_flow_api_with_see: before invoking a sub-flow via API, the node reads FlowOutputMode from the variable pool's system params; if it is None the node cannot know how the sub-flow returns its output and aborts. It means the flow's output-mode parameter was never set in the execution context for this flow_id.

Solutions

1. Open the flow configuration and explicitly set its output mode (e.g. streaming vs. message-collection) and re-publish.
2. If invoking programmatically, set variable_pool.system_params[ParamKey.FlowOutputMode] before calling the flow node.
3. Update old flow definitions by re-saving them so the output-mode field gets a default.
4. Add validation at flow publish time to reject flows missing output mode, turning this into an earlier schema error.

Example fix

# before: invoking flow without output mode
await flow_node.async_execute(variable_pool)
# after: ensure the param exists
from engine.params import ParamKey
variable_pool.system_params[ParamKey.FlowOutputMode] = FlowOutputMode.MESSAGE_COLLECTION
await flow_node.async_execute(variable_pool)

output_mode = variable_pool.system_params.get(ParamKey.FlowOutputMode, node_id=flow_node.node_id)
if output_mode is None:
    raise ValueError("FlowOutputMode must be set before invoking a flow node")

try:
    output = await flow_node.async_execute(...)
except CustomException as e:
    if "Flow output mode not configured" in str(e.cause_error):
        logger.error("set the flow's output mode in config and re-publish")

When it happens

Trigger: Calling a sub-flow node when the parent execution did not populate ParamKey.FlowOutputMode in variable_pool.system_params — e.g. direct/debug invocation of the flow, a missing default when the flow was published without output-mode configuration, or programmatic invocation that skipped parameter injection.

Common situations: API/debug runs that bypass the normal publish pipeline; older flow definitions created before the output-mode setting existed; manual construction of system_params in tests or scripts omitting the key.

Understand the failure class

Background: "is required", "must be set", "missing required field": configuration validation errors across open-source libraries — this error's family across 36 libraries.

Related errors

• 40028
• ENG_PROTOCOL_VALIDATE_ERROR
• 21601
• 21602
• 23900

AI-assisted analysis of iflytek/astron-agent@5e758547a8 (2026-09-12). Data as JSON: /api/errors/93290839539871a9. Report an issue: GitHub.

Appendix: source

Thrown at core/workflow/engine/nodes/flow/flow_node.py:199

        This method establishes a streaming connection to the target workflow
        and processes the response in real-time, handling different output modes
        and streaming content to dependent nodes when necessary.

        :param url: SSE endpoint URL for the workflow API
        :param inputs: Input parameters for the target workflow
        :param variable_pool: Variable pool for workflow context
        :param span: Tracing span for observability
        :param msg_or_end_node_deps: Message dependencies for streaming output
        :param event_log_node_trace: Optional node trace logging
        :return: Tuple containing (outputs_dict, token_usage_dict)
        :raises CustomException: When workflow execution fails or times out
        """
        # Get the output mode configuration for the flow
        output_mode = variable_pool.system_params.get(
            ParamKey.FlowOutputMode, node_id=self.node_id
        )
        if output_mode is None:
            raise CustomException(
                err_code=CodeEnum.WORKFLOW_EXECUTION_ERROR,
                cause_error=f"Flow output mode not configured for flow_id: {self.flowId}",
            )

        # Assemble request headers and body
        headers, req_body = await self._assemble_request(
            url, inputs, variable_pool, span, event_log_node_trace
        )

        # Initialize response containers
        outputs = {}
        token_usage = {}

        try:
            # Initialize content accumulators for streaming response
            result_content = ""
            result_reasoning_content = ""

View on GitHub (pinned to 5e758547a8)