Iteration Steps

Iteration steps provide control flow primitives for loops, conditionals, and sequence generation in Jetty workflows.

Available Steps

Step	Activity Name	Description
Number Sequence Generator	number_sequence_generator	Generate sequences of numbers
Conditional Branch	conditional_branch	Execute conditional logic
Loop Counter	loop_counter	Maintain loop iteration state

---

Number Sequence Generator

Generates lists or tuples of numbers based on configurable rules.

Activity Name

number_sequence_generator

Overview

This step generates sequences of numbers in various patterns. It's useful for creating lists to iterate over, generating test data, or producing numeric ranges for parallel processing.

Configuration Parameters

Parameter	Type	Default	Description
mode	string	"range"	Generation mode: range, pairs, arithmetic, geometric
start	int	0	Starting value
end	int	10	Ending value
step	int	1	Step increment (or ratio for geometric)
output_format	string	"list"	Output format: list, tuples, json
max_terms	int	100	Maximum terms for geometric sequences

Path Expression Parameters

Parameter	Description
start_path	Path to dynamic start value
end_path	Path to dynamic end value
step_path	Path to dynamic step value

Generation Modes

Range Mode

Simple range: [start, start+step, start+2*step, ..., end]

{
  "activity": "number_sequence_generator",
  "mode": "range",
  "start": 0,
  "end": 10,
  "step": 2
}
// Output: [0, 2, 4, 6, 8, 10]

Pairs Mode

Consecutive pairs: [(a, a+1), (a+1, a+2), ...]

{
  "activity": "number_sequence_generator",
  "mode": "pairs",
  "start": 0,
  "end": 6,
  "step": 2
}
// Output: [(0, 2), (2, 4), (4, 6)]

Arithmetic Mode

Arithmetic sequence with custom step.

{
  "activity": "number_sequence_generator",
  "mode": "arithmetic",
  "start": 5,
  "end": 20,
  "step": 3
}
// Output: [5, 8, 11, 14, 17, 20]

Geometric Mode

Geometric sequence with custom ratio.

{
  "activity": "number_sequence_generator",
  "mode": "geometric",
  "start": 1,
  "end": 100,
  "step": 2,
  "max_terms": 10
}
// Output: [1, 2, 4, 8, 16, 32, 64]

Examples

Generate Batch Indices

{
  "name": "create_batches",
  "activity": "number_sequence_generator",
  "config": {
    "mode": "range",
    "start": 0,
    "end_path": "data_loader.outputs.total_items",
    "step": 100
  }
}

Create Test Thresholds

{
  "name": "thresholds",
  "activity": "number_sequence_generator",
  "config": {
    "mode": "arithmetic",
    "start": 0,
    "end": 100,
    "step": 10,
    "output_format": "list"
  }
}

Output Structure

{
  "outputs": {
    "sequence": [0, 2, 4, 6, 8, 10],
    "stats": {
      "count": 6,
      "min": 0,
      "max": 10,
      "sum": 30,
      "avg": 5.0
    },
    "mode": "range",
    "parameters": {
      "start": 0,
      "end": 10,
      "step": 2,
      "output_format": "list"
    }
  }
}

---

Conditional Branch

Executes conditional logic based on input values.

Activity Name

conditional_branch

Overview

This step evaluates a condition and returns different outputs based on the result. Use it to implement branching logic within workflows.

Configuration Parameters

Parameter	Type	Default	Description
value_path	string	"init_params.value"	Path to the value to evaluate
condition_type	string	"equals"	Type of comparison
condition_value	any	null	Value to compare against
true_output	object	{"result": true}	Output when condition is true
false_output	object	{"result": false}	Output when condition is false

Path Expression Parameters

Parameter	Description
condition_value_path	Path to dynamic comparison value

Condition Types

Type	Description	Example
equals	Exact equality	value == condition_value
not_equals	Not equal	value != condition_value
greater_than	Greater than	value > condition_value
less_than	Less than	value < condition_value
greater_equal	Greater or equal	value >= condition_value
less_equal	Less or equal	value <= condition_value
contains	String contains	condition_value in value
exists	Value is not None	value is not None
is_empty	Value is empty/falsy	not value

Examples

Quality Gate

{
  "name": "quality_check",
  "activity": "conditional_branch",
  "config": {
    "value_path": "evaluator.outputs.score",
    "condition_type": "greater_equal",
    "condition_value": 0.8,
    "true_output": {
      "status": "passed",
      "action": "publish"
    },
    "false_output": {
      "status": "failed",
      "action": "review"
    }
  }
}

Check for Errors

{
  "name": "error_check",
  "activity": "conditional_branch",
  "config": {
    "value_path": "processor.outputs.error",
    "condition_type": "exists",
    "true_output": {
      "has_error": true,
      "next_step": "error_handler"
    },
    "false_output": {
      "has_error": false,
      "next_step": "continue"
    }
  }
}

Category Routing

{
  "name": "route_by_type",
  "activity": "conditional_branch",
  "config": {
    "value_path": "classifier.outputs.category",
    "condition_type": "equals",
    "condition_value": "urgent",
    "true_output": {
      "priority": "high",
      "queue": "urgent_queue"
    },
    "false_output": {
      "priority": "normal",
      "queue": "standard_queue"
    }
  }
}

Output Structure

{
  "outputs": {
    "status": "passed",
    "action": "publish",
    "condition_result": true,
    "evaluated_value": 0.92,
    "condition": {
      "type": "greater_equal",
      "expected": 0.8,
      "actual": 0.92,
      "result": true
    }
  }
}

---

Loop Counter

Maintains loop counter and iteration state.

Activity Name

loop_counter

Overview

This step manages iteration state for loop-based workflows. It tracks the current iteration, calculates progress, and determines when the loop should complete.

Configuration Parameters

Parameter	Type	Default	Description
start_count	int	0	Initial count if no current count found
max_iterations	int	10	Maximum number of iterations
increment	int	1	Amount to increment each iteration

Path Expression Parameters

Parameter	Description
current_count_path	Path to current iteration count
max_iterations_path	Path to dynamic max iterations
increment_path	Path to dynamic increment value

Examples

Basic Loop Control

{
  "name": "loop_control",
  "activity": "loop_counter",
  "config": {
    "current_count_path": "init_params.iteration",
    "max_iterations": 5,
    "increment": 1
  }
}

Dynamic Loop Bounds

{
  "name": "dynamic_loop",
  "activity": "loop_counter",
  "config": {
    "current_count_path": "previous_iteration.outputs.next_count",
    "max_iterations_path": "data_loader.outputs.total_batches",
    "increment": 1
  }
}

Output Structure

{
  "outputs": {
    "current_count": 3,
    "next_count": 4,
    "is_complete": false,
    "continue_loop": true,
    "progress": 0.4,
    "remaining_iterations": 6,
    "loop_info": {
      "current": 3,
      "max": 10,
      "increment": 1,
      "progress_percent": 40.0
    }
  }
}

---

Common Patterns

Iterative Refinement Loop

Use loop counter with conditional to iteratively improve output:

{
  "steps": ["generate", "evaluate", "check_quality", "loop_control"],
  "step_configs": {
    "generate": {
      "activity": "litellm_chat",
      "prompt_path": "init_params.prompt",
      "model": "gpt-4o"
    },
    "evaluate": {
      "activity": "simple_judge",
      "instruction": "Rate the quality of this output",
      "judge_type": "scale",
      "scale_range": [1, 10],
      "item_path": "generate.outputs.text"
    },
    "check_quality": {
      "activity": "conditional_branch",
      "value_path": "evaluate.outputs.results[0].score",
      "condition_type": "greater_equal",
      "condition_value": 8,
      "true_output": {"done": true},
      "false_output": {"done": false}
    },
    "loop_control": {
      "activity": "loop_counter",
      "max_iterations": 3
    }
  }
}

Batch Processing with Sequences

Generate batch boundaries for large dataset processing:

{
  "steps": ["count_items", "generate_batches", "process_batches"],
  "step_configs": {
    "count_items": {
      "activity": "read_text_file",
      "text_path": "init_params.data_file"
    },
    "generate_batches": {
      "activity": "number_sequence_generator",
      "mode": "range",
      "start": 0,
      "end_path": "count_items.outputs.line_count",
      "step": 100
    },
    "process_batches": {
      "activity": "list_emit_await",
      "items_path": "generate_batches.outputs.sequence",
      "task_reference": {"task_name": "process_batch"},
      "data_mapping": {
        "offset": "{{ $item }}",
        "limit": 100
      }
    }
  }
}

Threshold Testing

Test multiple thresholds to find optimal values:

{
  "steps": ["generate_thresholds", "test_all", "find_best"],
  "step_configs": {
    "generate_thresholds": {
      "activity": "number_sequence_generator",
      "mode": "arithmetic",
      "start": 50,
      "end": 100,
      "step": 5
    },
    "test_all": {
      "activity": "list_emit_await",
      "items_path": "generate_thresholds.outputs.sequence",
      "task_reference": {"task_name": "test_threshold"},
      "data_mapping": {
        "threshold": "{{ $item }}"
      }
    },
    "find_best": {
      "activity": "extract_from_trajectories",
      "trajectory_list_path": "test_all.outputs.trajectory_references",
      "extract_keys": {
        "threshold": "init_params.threshold",
        "accuracy": "steps.evaluate.outputs.accuracy"
      }
    }
  }
}

Best Practices

Sequence Generation

• Use appropriate modes for your use case
• Set max_terms for geometric sequences to prevent runaway generation
• Consider memory usage with very large sequences

Conditional Logic

• Keep conditions simple and readable
• Use descriptive true_output and false_output values
• Consider using multiple conditions for complex routing

Loop Control

• Always set reasonable max_iterations to prevent infinite loops
• Track progress for long-running iterative processes
• Use is_complete to determine loop termination

Related Steps

• List Emit Await - Process sequences in parallel
• Simple Judge - Evaluate loop outputs
• LiteLLM Chat - Generate content in loops