The Loop node is a component in the Sequential Agents category, designed to create loops within agent workflows. It allows the flow to return to a specific previous node, enabling repetitive or conditional execution of parts of the sequence.
Name: seqLoop
Label: Loop
Version: 2.0
Type: Loop
Category: Sequential Agents
Base Class: Loop
The Loop node accepts the following input parameters:
Agent | Condition | LLM | Tool Node
Name: sequentialNode
Type: Agent | Condition | LLMNode | ToolNode
Description: The predecessor node(s) in the sequence
Required: Yes
List: Yes (can accept multiple nodes)
Loop To
Name: loopToName
Type: string
Description: Name of the agent/LLM to loop back to
Required: Yes
Placeholder: “Agent”
The node requires at least one predecessor node (Agent, Condition, LLM, or Tool).
The name of the node to loop back to must be specified.
The node’s output is hidden (hideOutput: true).
It returns an ISeqAgentNode object containing:
id: The node’s ID
node: The name of the node to loop to (lowercase, underscores instead of spaces)
name: Same as ‘node’
label: The original label of the node to loop to
type: Always set to ‘agent’
predecessorAgents: Array of input sequential nodes
output: Same as ‘node’
The Loop node is used to create cycles in sequential agent workflows. It’s particularly useful for:
Implementing iterative processes where certain steps need to be repeated.
Creating conditional loops where the workflow returns to a previous state based on certain conditions.
Building complex decision trees or state machines within agent-based systems.
To use the Loop node:
Connect it to one or more predecessor nodes in the sequence.
Specify the name of the node you want to loop back to in the “Loop To” field.
The node will then create a loop in the workflow, directing the execution back to the specified node when reached.
The node will throw an error if:
No predecessor nodes are provided
The “Loop To” name is not specified