App builders · guidance updated 15 Nov 2026
Bolt.new prompt guide
Plan Mode now includes web research for real-time information, quick action buttons for immediate execution, and read-only database access. Use Plan Mode from homepage to plan before building. Standard agent for typical tasks; Max for complex work only.
How to prompt Bolt.new
- Use Plan Mode from homepage or mid-project to plan, debug, research, and explore before building—it uses fewer tokens and includes web research.
- Start with clear app purpose, target users, and core features as a blueprint in your first prompt.
- Specify tech stack upfront: framework, styling, backend, database, and integrations.
- Build incrementally: create core pages and navigation first, then add features one at a time.
- Make one change per prompt, verify it works, then move to the next feature.
- Name exact files, components, classes, or functions and explicitly state what must not change.
- Use Target or Lock files in Code view to focus Bolt and prevent unintended edits.
- Add error handling and logging from the start, and use Plan Mode to troubleshoot rather than repeated fix attempts.
- Describe UI concretely: layout, palette, components, and responsive behavior.
- Use quick action buttons in Plan Mode (Implement, Show example, Refine) to execute suggestions efficiently.
- Choose Standard agent for typical tasks; use Max agent only for complex refactors or tasks requiring deep reasoning.
- Enable connectors only when actively needed to minimize token consumption and context bloat.
- Use built-in buttons (Publish, Version History) instead of prompting for these tasks to avoid token waste.
The shape of a good prompt
[App purpose and users] [Tech stack: framework, styling, backend, integrations] [Core features for v1, max 3] [This step: specific feature to build now] [UI and design: layout, palette, components, responsive] [Data model] [Error handling and logging] [Out of scope / don't change]
Avoid
- Long feature lists and multiple requirements in a single prompt
- Vague edits that don't name files or specific elements
- Repeated 'Attempt fix' clicks in error loops—use Plan Mode instead to troubleshoot
- Prompting for tasks built-in buttons already handle (publish, version history, revert)
- Letting chat context grow huge without clearing it or duplicating the project
- Enabling connectors unless actively using them
- Using Max agent for straightforward or simple tasks
- Skipping error handling setup—add logging from the start to prevent token-burning fix loops
Example
Request a habit tracker app
Build a habit tracker web app for individuals who want to build 3-5 daily habits. Stack: React + Vite + TypeScript, Tailwind CSS, Supabase for auth and data. Step 1 only: project setup, email/password login, and a 'Today' screen listing my habits with a checkbox each. Don't build stats or reminders yet. Data: habits (id, user_id, name, colour, created_at), completions (habit_id, date). UI: single-column mobile-first layout, white cards on a light grey background, one accent colour per habit, large tap targets, an empty state with an 'Add your first habit' button. Add clear error messages and console logging around all Supabase calls.