​
Overview

The end_step function completes an active step, optionally updating its final state and evaluation. This marks the step as finished and allows you to create a new step.

​
Syntax

lai.end_step(
    step_id: Optional[str] = None,
    state: Optional[str] = None,
    action: Optional[str] = None,
    goal: Optional[str] = None,
    eval_score: Optional[float] = None,
    eval_description: Optional[str] = None,
    screenshot: Optional[str] = None,
    screenshot_path: Optional[str] = None
) -> None

​
Parameters

​
step_id
string
The ID of the step to end. If not provided, ends the current active step.
​
state
string
Final state description to set before ending the step.
​
action
string
Final action description to set before ending the step.
​
goal
string
Final goal description to set before ending the step.
​
eval_score
float
Final evaluation score (typically 0.0 to 1.0) indicating step success/quality.
​
eval_description
string
Final evaluation description explaining the outcome.
​
screenshot
string
Final base64-encoded screenshot. Provide either this or screenshot_path.
​
screenshot_path
string
Path to a final screenshot file. Provide either this or screenshot.

​
Returns

This function returns None.

​
Examples

​
Basic Usage

import lucidicai as lai

lai.init(session_name="Simple Workflow")

# Create and end a step
lai.create_step(
    state="Processing request",
    action="Generate response"
)

# Do work...
result = process_request()

# End the step
lai.end_step()

​
With Final Evaluation

import lucidicai as lai

lai.init(session_name="Task Execution")

lai.create_step(
    state="Starting complex task",
    action="Execute multi-stage process",
    goal="Complete all stages successfully"
)

try:
    result = execute_complex_task()
    
    # End with success
    lai.end_step(
        eval_score=1.0,
        eval_description="All stages completed successfully",
        state="Task completed with result: " + str(result)
    )
except Exception as e:
    # End with failure
    lai.end_step(
        eval_score=0.0,
        eval_description=f"Task failed: {str(e)}",
        state="Error encountered during execution"
    )

​
Multi-Step Workflow

import lucidicai as lai

lai.init(session_name="Document Analysis")

# Step 1: Extract text
lai.create_step(state="PDF received", action="Extract text")
text = extract_pdf_text(pdf_path)
lai.end_step(
    eval_score=1.0 if text else 0.0,
    eval_description=f"Extracted {len(text)} characters"
)

# Step 2: Analyze content
lai.create_step(state="Text extracted", action="Analyze content")
analysis = analyze_text(text)
lai.end_step(
    eval_score=0.9,
    eval_description="Analysis completed with high confidence"
)

# Step 3: Generate report
lai.create_step(state="Analysis complete", action="Generate report")
report = generate_report(analysis)
lai.end_step(
    eval_score=1.0,
    eval_description="Report generated successfully",
    screenshot_path="/tmp/report_preview.png"
)

lai.end_session()

​
Conditional Ending

import lucidicai as lai

lai.init(session_name="Quality Check")

step_id = lai.create_step(
    state="Running quality checks",
    action="Validate output quality",
    goal="Ensure output meets standards"
)

# Run multiple checks
checks_passed = []
checks_passed.append(check_format(output))
checks_passed.append(check_content(output))
checks_passed.append(check_accuracy(output))

# End based on results
pass_rate = sum(checks_passed) / len(checks_passed)

if pass_rate == 1.0:
    lai.end_step(
        step_id=step_id,
        eval_score=1.0,
        eval_description="All quality checks passed"
    )
elif pass_rate >= 0.7:
    lai.end_step(
        step_id=step_id,
        eval_score=pass_rate,
        eval_description=f"{sum(checks_passed)}/{len(checks_passed)} checks passed"
    )
else:
    lai.end_step(
        step_id=step_id,
        eval_score=0.0,
        eval_description="Quality standards not met"
    )

​
Notes

  • A step must be active to be ended
  • Once ended, a step cannot be updated or modified
  • Only one step can be active at a time per session
  • Final parameters override any previous values
  • If no active step exists and step_id is not provided, raises an error

​
Error Handling

from lucidicai.errors import InvalidOperationError

try:
    lai.end_step()
except InvalidOperationError:
    print("No active step to end")

# Ending a specific step
try:
    lai.end_step(step_id="invalid-step-id")
except InvalidOperationError:
    print("Step not found or already ended")

​
Best Practices

  1. Always End Steps: Ensure every created step is properly ended
  2. Final Evaluation: Provide eval_score to indicate success/failure
  3. Descriptive Endings: Use eval_description to explain the outcome
  4. Error Handling: End steps even when errors occur
  5. Sequential Flow: End current step before creating a new one

​
See Also