diff --git a/tests/ci/test_sequential_agents_reuse.py b/tests/ci/test_sequential_agents_reuse.py index 5c8bea46a..44429aedd 100644 --- a/tests/ci/test_sequential_agents_reuse.py +++ b/tests/ci/test_sequential_agents_reuse.py @@ -171,45 +171,30 @@ class TestSequentialAgentsReuse: urls3 = [h.state.url for h in history3.history if h.state and h.state.url] assert httpserver.url_for('/page3') in urls3, f'Expected {httpserver.url_for("/page3")} in URLs: {urls3}' - # Agent 4: Extract content from page 3 + # Agent 4: Take a screenshot to confirm we're on page 3 agent4_actions = [ """{ - "thinking": "Extracting page content", + "thinking": "Taking screenshot on page 3", "evaluation_previous_goal": "Successfully navigated to page 3", - "memory": "On page 3, need to capture content", - "next_goal": "Extract page content", - "action": [ - {"extract_page_content": {}} - ] - }""", - """{ - "thinking": "Taking screenshot", - "evaluation_previous_goal": "Extracted content", - "memory": "Content extracted, taking screenshot", - "next_goal": "Take screenshot", + "memory": "On page 3, need to take screenshot", + "next_goal": "Take screenshot of page 3", "action": [ {"screenshot": {}} ] - }""", + }""" ] agent4 = Agent( - task='Extract content from page 3', + task='Take screenshot on page 3', llm=create_mock_llm(agent4_actions), browser_session=browser_session, ) - history4 = await agent4.run(max_steps=3) + history4 = await agent4.run(max_steps=2) assert len(history4.history) >= 1 - # Verify all agents completed successfully - found_success = False - for step in history4.history: - for result in step.result: - if hasattr(result, 'extracted_content') and result.extracted_content: - if 'Success!' in result.extracted_content: - found_success = True - break - assert found_success, "Did not find 'Success!' in extracted content" + # Verify that agent 4 was on page 3 + urls4 = [h.state.url for h in history4.history if h.state and h.state.url] + assert httpserver.url_for('/page3') in urls4, f'Expected {httpserver.url_for("/page3")} in URLs: {urls4}' # Clean up await browser_session.stop()