App builders · guidance updated 22 Nov 2026

Replit Agent prompt guide

Plan in plain language or use optional /plan shortcut. Replit recognizes when planning helps and proposes plans proactively. Build Mode is default; ask Replit to plan before implementing for complex work.

Free. We don't store what you type.

How to prompt Replit Agent

  1. State the outcome clearly: what you want to build, create, analyze, or decide.
  2. Add useful context: relevant files, examples, screenshots, integrations, web sources.
  3. Define important boundaries: what to keep, exclude, prioritize.
  4. Explain how you'll use the result: audience, decision, format, next action.
  5. Break complex requests into bullet points using clear, straightforward language.
  6. Use positive language: describe what you want, not what to avoid.
  7. Ask Replit to plan in plain language for complex tasks before implementation.
  8. Be specific: define fields, validation, formats, edge cases, and expected behavior.
  9. Provide relevant files and examples instead of entire projects.
  10. For debugging, include exact error messages, file names, code context, and what you've tried.
  11. Build incrementally using checkpoints after each step; roll back if needed.
  12. Refine iteratively through follow-ups, adding detail or simplifying until results match your needs.

The shape of a good prompt

[Outcome: what to build/create/analyze and for whom] [Useful context: files, examples, screenshots, integrations, web sources] [Boundaries: what to preserve, prioritize, exclude] [Format or use: audience, decision, next action] [Follow-ups: add detail, iterate, refine toward result]

Avoid

  • Vague descriptions without specifics (fields, behavior, formats, edge cases)
  • Negative phrasing instead of positive direction ('handle 10k rows fast', not 'don't be slow')
  • Attaching entire projects instead of mentioning specific files
  • Bug reports without exact error messages, file names, or code context
  • Requesting a complete platform in one prompt without breaking into stages
  • Mixing unrelated tasks in one chat
  • Assuming your first prompt is perfect; iterate if results aren't right

Example

Request an expense tracker for my team

Build a web app where a small team (5-15 people) logs work expenses and a manager approves them.

User flow: employee logs in -> submits expense (date, amount in GBP, category, description, receipt photo) -> manager sees a pending list -> approves or rejects with a comment -> employee sees the status.

Stage 1 only: login, the submit form, and the employee's own expense list. Approval comes next.

Requirements:
- Amount must be a positive number with 2 decimals
- Categories: Travel, Meals, Equipment, Other
- Receipt upload accepts JPG/PNG/PDF up to 5 MB
- List sorted newest first, with status badges (Pending/Approved/Rejected)

Use a PostgreSQL database. Plan the data model first and show me the task list before building.

The Weekly Tally

Useful updates, insights and no-hype advice. Read this week's digest.

One email a week. Unsubscribe in one click.