Node Error Handling Strategy

All major GoInsight.AI nodes include an error handling strategy designed to manage node errors. When a critical error prevents output during node execution, this mechanism activates to avoid interrupting the entire workflow.

The following nodes currently support the "Error Handling Strategy" option; more will be added in the future:

Agent
Code
HTTP Request
Knowledge Base Retrieval
Document Read
Document Write

LLM
Question Classifier
Variable Extractor
Template
Json Variable Extractor
Tool Call

Configuration Method

To configure error handling, select a supported node and locate the Error Handling drop-down menu in the configuration panel. The default option is Do Not Handle. The available settings are:

None
Report an error immediately when node execution fails, then terminate the workflow.
Default Value
If a node encounters an execution error, the workflow continues running. All output variables of that node will be replaced with specified default values. Different nodes require different default values, which you will be prompted to fill in based on the expected output variables.

Node	Default Value Configuration Required When Node Errors Occur
Agent	Variable name: Text variable type: String, optional; outputs an empty string if left blank
LLM	Variable name: Text variable type: String, optional; outputs an empty string if left blank
Knowledge Base Retrieval	Variable name: Result variable type: Array[Object], optional; ignored if left blank
Document Read	Variable name: User-defined output variable name for the document Variable type: String, optional; outputs an empty string if left blank. Each document can have its own error default value.
Document Write	Has an error handling option but no default value setting
HTTP Request	Outputs three fixed variables: 1. StatusCode: status code, variable type: Number 2. Headers, variable type: Object 3. Body, variable type: String All optional; if left blank, variables are ignored or output as empty strings.
Question Classifier	Variable name: ClassName variable type: String, optional; outputs an empty string if left blank
Template	Variable name: Output variable type: String, optional; outputs an empty string if left blank
Code	Variable name: User-configured output variable name for the code execution node Variable type: Same as the output variable type configured for the code execution node All optional; if left blank, the variable will be ignored or output as an empty string
JSON Variable Extractor	Variable name: User-configured output variable name for the JSON extractor node Variable type: Corresponding variable type configured for the JSON variable extractor node All optional; if left blank, the variable will be ignored or outputs as an empty string
Variable Extractor	Outputs an additional variable beyond user-defined variables: Additional output variable: Variable name: $AllFound Variable type: Bool, optional; ignored if left blank. Indicates whether AI variable extraction succeeded—true means success, false means failure. Variable name and type vary based on user's "Extract Variables" settings: Variable name: User-configured output variable name for the natural semantic variable extractor node Variable type: Corresponding variable type configured for the natural semantic variable extractor node All optional, if left blank, variable will be ignored or output as an empty string

Error Handling Branch
Customize error handling branch logic within the workflow designer. By connecting these branches, you can define responses to node execution errors, such as sending warnings, running backup tasks, or logging events.

http request error redirect to end branch

As shown in the above diagram, when an error occurs in the HTTP Request node, the workflow is directed to the End (mistake) branch.

Error Handling Strategy for Iteration

Unlike the other nodes mentioned above, iteration nodes typically process a large number of tasks, and errors may occur when handling individual elements. To prevent a single element's error from interrupting all tasks, iteration nodes provide a separate error handling strategy. You can configure the response method for exceptions under error handling options. There are two choices:

Terminate
If an abnormal output is detected, terminate this node and output an error message.
Remove Error Output
Ignore exception information and continue processing the remaining elements. The output will only contain correct information.

Users can select the error handling method that best fits their loop design requirements.

Updated on: Aug 26, 2025

No Results