ContextQMD
Back to Copilotkit

QA: Human in the Loop — Claude Agent SDK (Python)

showcase/integrations/claude-sdk-python/qa/hitl-in-chat.md

1.57.02.2 KB
Original Source

QA: Human in the Loop — Claude Agent SDK (Python)

Prerequisites

  • Demo is deployed and accessible
  • Agent backend is healthy (check /api/health)

Test Steps

1. Basic Functionality

  • Navigate to the HITL demo page (/demos/hitl) > Note: the URL path /demos/hitl is intentionally shorter than the feature id hitl-in-chat.
  • Verify the chat interface loads in a centered max-w-4xl container
  • Verify the chat input placeholder "Type a message" is visible
  • Send a basic message
  • Verify the agent responds

2. Feature-Specific Checks

Suggestions

  • Verify "Simple plan" suggestion button is visible
  • Verify "Complex plan" suggestion button is visible
  • Click the "Simple plan" suggestion
  • Verify it triggers a message about planning a trip to Mars in 5 steps

Step Selection and Approval

The HITL demo renders a single StepSelector card regardless of whether the underlying flow uses an interrupt hook or a frontend HITL tool. The framework-specific hook name is an implementation detail covered in each package's demo source; QA below validates the user-visible card behavior.

  • Send "Plan a trip to Mars in 5 steps"
  • Verify the StepSelector card appears (data-testid="select-steps")
  • Verify step items are displayed with checkboxes (data-testid="step-item")
  • Verify step text is visible (data-testid="step-text")
  • Verify the selected count display shows "N/N selected"
  • Toggle a step checkbox off and verify the count decreases
  • Toggle it back on and verify the count increases
  • Click "Perform Steps (N)" / "Confirm (N)" button
  • Verify the agent continues processing after confirmation
  • In a new conversation, trigger the same flow and click "Reject" (where present)
  • Verify the card reflects the decision (Accepted / Rejected) and buttons disable

3. Error Handling

  • Send an empty message (should be handled gracefully)
  • Verify no console errors during normal usage

Expected Results

  • Chat loads within 3 seconds
  • Agent responds within 10 seconds
  • Step selector renders with toggleable checkboxes
  • Accept/Reject flow completes without errors
  • No UI errors or broken layouts