Agent Mode
AI Agent · 6 articles
Can I see what Lovable is doing while it works?
Yes, you can watch Agent mode execute in real-time. See your code update, preview render, and logs populate as it works. This transparency helps you understand what's being built.
Can multiple messages be sent while working in Agent mode?
No, Agent mode processes one request at a time. Wait for the current request to complete before sending a new prompt.
Does Lovable modify my project directly when in Agent mode?
Yes, Agent mode implements and verifies changes directly in your project. It thinks, plans, and takes autonomous actions to build features end-to-end. Use Plan mode first if you want to reason through changes before execution.
How do I stop a request, and what happens when I do?
You can stop an in-progress Agent mode request by clicking Stop. Any changes made up until that point are saved, and you can revise or continue from there.
Review knowledge periodically
Review your project and workspace knowledge every few months or after major changes. If Lovable is making decisions you don't like, check if your knowledge is outdated or unclear. Update it to reflect current preferences, new standards, or lessons learned. Think of knowledge as living documentation that evolves with your project.
What happens if errors occur during implementation?
Lovable detects errors during implementation and uses the free 'Try to fix' button to auto-repair issues (~60% success rate for simple glitches). Use the Cloud → Logs tab to debug complex issues, or revert and re-prompt with a clearer description.
