Docs / Strand / advanced/infinite-loops

Infinite Loop Prevention

Strand automatically prevents infinite loops when workflows call each other.

How It Works

The system tracks a call chain - a list of workflow IDs in the current execution path. Before executing a nested workflow, it checks if that workflow is already in the chain.

Example

code


Workflow A calls Workflow B
  Call chain: [A]
  
Workflow B calls Workflow A
  Call chain: [A, B]
  
System detects: A is already in chain → ERROR

Error Message

When an infinite loop is detected, you'll see:

code


Infinite loop detected: workflow 'workflow-id' is calling itself 
(directly or indirectly). Call chain: workflow-a -> workflow-b -> workflow-a

Direct Loops

A workflow calling itself directly:

code


Workflow A → Flow Call → Workflow A

This is detected immediately when the Flow Call node tries to execute.

Indirect Loops

A workflow calling itself through other workflows:

code


Workflow A → Workflow B → Workflow C → Workflow A

This is detected when Workflow C tries to call Workflow A, which is already in the chain.

Prevention Tips

Design workflows carefully

Plan the workflow hierarchy before building

Use descriptive names

Makes it easier to spot potential loops

Document dependencies

Keep track of which workflows call which others

Test incrementally

Test nested workflows before combining them

Valid Patterns

Sequential Calls

code


A → B → C

✅ Valid - no loops

Parallel Calls

code


A → B
A → C

✅ Valid - both B and C are called from A, but don't call each other

Conditional Calls

code


A → (if condition) → B
A → (else) → C

✅ Valid - only one path executes

Debugging

If you encounter an infinite loop error:

Check the call chain in the error message
Review your workflow structure
Look for circular dependencies
Consider refactoring to break the cycle

← Back to Strand docs