The extract content in browser use fails

[Mohit-Dhawan98]

Bug Description

🔧 Tool arguments: {"action": "extract_content", "goal": "Extract the complete list of countries and their capitals from this table"}
2025-05-22 23:21:14.090 | INFO | app.agent.toolcall:execute_tool:217 - 🔧 Activating tool: 'browser_use'...
2025-05-22 23:21:14.595 | ERROR | app.llm:ask_tool:868 - OpenAI API error: Error code: 400 - {'error': {'code': 'invalid_request_error', 'message': 'At least one non-system, non-developer message is required.', 'type': 'invalid_request_error', 'param': None}}
2025-05-22 23:21:14.595 | ERROR | app.llm:ask_tool:874 - API error: Error code: 400 - {'error': {'code': 'invalid_request_error', 'message': 'At least one non-system, non-developer message is required.', 'type': 'invalid_request_error', 'param': None}}

Bug solved method

Environment information

• System version: Mac M3
• Python version: python 3.11
• OpenManus version or branch:
• Installation method (e.g., pip install -r requirements.txt or pip install -e .): uv pip install -r requirements.txt

Extra information

No response

[broimsuperman]

Same issue here!

[Mohit-Dhawan98]

fixed it by adding the following line in browser_use_tool.py in line 393

                messages = [
                    {"role": "system", "content": prompt},
                    {"role": "user", "content": "Extract the relevant content from this page based on the goal."}
                ]

[github-actions]

This issue has been inactive for 30 days. Please comment if you have updates.