Skip to main content
Choose a modeLovable has two modes:
  • Plan mode → think through the problem, explore options, and decide on an approach
  • Agent mode → implement changes and verify the outcome
Plan mode is for decision-making. Agent mode is for execution.The two modes are designed to work together, and you can switch between them at any time.

​
Overview

Agent mode is Lovable’s autonomous execution mode, designed to implement changes directly in your project. When you give Lovable a task, it takes ownership of execution end to end. It understands your intent, explores the codebase for context, applies changes across files, and resolves issues that appear during development. While it works, you can see progress through visible tasks and review the results before moving on.

​
What Agent mode is for

You can use Agent mode to:
  • Implement features or changes once an approach is decided
  • Fix bugs or errors end to end
  • Refactor code across multiple files
  • Apply coordinated changes across frontend, backend, and configuration
  • Debug issues that arise during implementation
  • Inspect logs and network activity to identify problems
  • Fetch external documentation or assets when needed
  • Generate or edit images and videos for use in your project
  • Verify results before finishing
All changes are applied directly to your project and surfaced through file diffs and summaries. You can follow along step by step in the Details view.

​
How to use Agent mode

Agent mode is active unless you switch to Plan mode. To use it, describe what you want and send your message. Clear requirements and constraints help Lovable produce better results. Be specific about what you want 
Add a contact form with name, email, and message fields. Validate email format and show error messages inline.
Describe expected behavior 
When users click ‘Add to Cart’, show a success message and update the cart count in the header. Store cart items in local storage.
Provide context for complex changes 
Update the dashboard to show sales data from the last 30 days. Use the existing Chart component and match the styling used on the analytics page.
Set guardrails for sensitive areas 
Add a new feature to @src/pages/dashboard. Do not modify @src/shared/Layout.tsx or the existing authentication logic.

​
Tasks and execution visibility

While Lovable is working, tasks appear in the chat interface showing:
  • Current step being executed
  • Files being modified
  • Tools being used (search, web fetch, image generation)
  • Progress through multi-step implementations
This visibility helps you:
  • Follow progress on complex builds
  • Understand the steps being performed
  • Stay oriented during complex changes
  • Spot issues early if something seems off

​
Message queue

Lovable processes one task at a time. If you send additional messages while it is working, or multiple collaborators send messages simultaneously:
  • They are automatically queued
  • The queue is visible above the chat input
  • They are processed in order
You can reorder, pause, or remove queued messages at any time.

​
Debugging and verification

When using Agent mode, Lovable does more than ensure code compiles. As part of investigation and problem solving, the agent can:
  • observe build errors and failures from test runs
  • inspect console output and network requests when verification tools are used
  • reproduce reported issues using browser testing or targeted backend calls
  • verify that fixes work as expected by running appropriate checks
Lovable has access to verification tools such as browser testing, frontend tests, and edge function verification to help understand system behavior, validate user workflows, and ensure backend logic works as expected. Most of these tools run only when you ask for them. For a detailed explanation of testing and verification capabilities, see Test and verify your app.

​
Pricing

Pricing for Agent mode is usage-based. Cost depends on factors such as:
  • Number of files modified
  • Complexity of logic changes
  • Amount of codebase exploration required
  • Use of tools such as verification, browser checks, web search, or image generation
Many requests cost less than one credit, while more complex tasks may cost more. You can view the cost of a message from the three-dot menu on that message. See credit usage for more information.

​
FAQ

Yes. Lovable applies changes directly to your project when using Agent mode. All modifications are visible through file diffs and summaries.
Lovable can inspect logs, runtime output, and network activity and iterate on fixes until the issue is resolved or clarified. See Test and verify your app for more information.
Yes. Visible tasks show each step of execution, including progress and which files are being modified.
Yes. Additional messages are queued and processed in order. You can reorder, pause, or remove queued messages at any time.
Click the stop button while Lovable is responding. The current task will halt and any uncommitted changes will not be saved.
Switch to Plan mode when you want to:
  • Think through a complex change before implementation
  • Compare multiple architectural approaches
  • Review or edit a plan before any code is changed
  • Investigate an issue and decide on a solution first