Replit Docs

Agent automatically creates this file in your project’s root directory using proven best practices. Agent includes its contents in the context to help it understand your preferences, project structure, and coding style.

Agent using replit.md to build a project

How `replit.md` works

Agent first creates a replit.md file in your project’s root directory using proven best practices. This file includes:

Basic project information
Recommended coding patterns
Common preferences for your project type

When Agent processes your requests, it automatically reads your replit.md file and uses its contents to:

Understand your project’s architecture and conventions
Follow your preferred coding patterns and style
Use your specified package managers and dependencies

Agent can also update your replit.md file as it learns more about your project and makes changes to your application.

You can edit the replit.md file to customize Agent’s behavior.

Setting up `replit.md`

Automatic generation

When you create a new project with Agent, it automatically generates a replit.md file using proven best practices. This file appears in your project’s root directory and includes:

Basic project information
Recommended coding patterns
Common preferences for your project type

Manual creation

You can create your own replit.md file by adding a new file named replit.md in your project’s root directory. Agent will automatically detect and use this file in future conversations.

replit.md must be located in your project’s root directory to work properly.

Regenerating `replit.md`

If your replit.md file becomes corrupted or you want to start fresh:

Delete the existing replit.md file from your project root
Start a new conversation with Agent
Agent will automatically generate a new replit.md file based on your current project

Best practices

Be specific and clear

Write clear, specific instructions that help Agent understand exactly what you want:

## Coding Style

- Use TypeScript for all new JavaScript files
- Prefer functional components with hooks over class components
- Use Tailwind CSS for styling, avoid inline styles
- Always include TypeScript types for function parameters and return values

For a more detailed guide on prompt formatting, check out the Anthropic prompt engineering guide.

Use examples

Examples help Agent understand your preferences better than abstract descriptions:

## API Error Handling

When creating API endpoints, always use this error handling pattern:

```javascript
app.get('/api/users', async (req, res) => {
  try {
    const users = await getUsersFromDatabase();
    res.json({ success: true, data: users });
  } catch (error) {
    console.error('Error fetching users:', error);
    res.status(500).json({ 
      success: false, 
      error: 'Failed to fetch users' 
    });
  }
});

Define communication preferences

Tell Agent how you prefer to receive information and updates:

## Communication Style

- Before implementing changes, explain what you're going to do and why
- Break down complex tasks into clear steps
- Ask for clarification if requirements are unclear
- Provide brief explanations for technical decisions

Specify project context

Help Agent understand your project’s purpose and constraints:

## Project Context

This is a social media app for book lovers where users can:
- Create reading lists and share book recommendations
- Follow other readers and see their updates
- Rate and review books they've read

Target audience: Casual readers aged 25-45
Tech stack: React frontend, Node.js backend, PostgreSQL database

Example `replit.md` configurations

Web application project

# MyApp Project Guidelines

## Project Overview
A task management web application built with React and Express.js.

## Technology Preferences
- Frontend: React with TypeScript, Tailwind CSS
- Backend: Express.js with TypeScript
- Database: Neon PostgreSQL with Drizzle ORM
- Package Manager: npm (not yarn or pnpm)

## Coding Standards
- Use functional components with hooks
- Implement proper error boundaries
- Follow REST API conventions for endpoints
- Use descriptive variable and function names

## Communication Style
- Explain your approach before making changes
- Provide code comments for complex logic
- Ask questions if requirements are ambiguous

Data analysis project

# Data Analysis Project Guidelines

## Project Context
Analyzing customer behavior data to identify trends and insights.

## Technology Preferences
- Python with Streamlit for interactive web applications
- pandas, numpy, and matplotlib for data analysis

## Analysis Standards
- Include clear documentation for all analysis steps
- Create interactive visualizations with Streamlit components
- Validate data quality before analysis
- Export results to CSV format and display in Streamlit tables

## Communication Style
- Explain statistical methods and assumptions
- Provide context for findings and recommendations
- Use clear, non-technical language for business insights
- Create interactive dashboards for stakeholder presentations

API development project

# API Development Guidelines

## Project Overview
Building a RESTful API for a mobile app backend.

## Technology Stack
- Node.js with Express.js
- MongoDB with Mongoose
- JWT for authentication
- Package manager: npm

## API Standards
- Use semantic HTTP status codes
- Implement proper input validation
- Include rate limiting on public endpoints
- Follow OpenAPI specification for documentation

## Security Requirements
- Validate all user inputs
- Use environment variables for sensitive data
- Implement proper CORS policies
- Include request logging for debugging

## Testing Approach
- Write unit tests for all endpoints
- Include integration tests for critical flows
- Use Jest as the testing framework

Advanced usage

Dynamic project guidance

Update your replit.md file as your project evolves to provide current context:

## Current Development Phase
The team is currently focused on implementing user authentication.
Priority features:
1. User registration and login
2. Password reset functionality  
3. Email verification

## Recent Decisions
- Switched from local authentication to OAuth with Google
- Added Redux for state management
- Decided to use Material-UI for consistent styling

Integration with external tools

Reference external documentation and tools in your replit.md:

## External Resources
- Design system: Follow the Figma design system at [link]
- API documentation: Reference our internal API docs
- Coding standards: Follow the team style guide in our wiki

## Deployment Process
- Test changes in development environment first
- Use the CI/CD pipeline for staging published apps
- Require code review before production releases

Combining web search with `replit.md`

Agent can perform web searches to find current information, libraries, and solutions. When combined with your replit.md file, you get both up-to-date knowledge and project-specific guidance.

Best practices for web search integration

Be specific about what you want Agent to research:

## Research Preferences
- Always check for the latest versions of dependencies before suggesting updates
- Research current best practices for security implementations
- Look up recent performance optimization techniques for our tech stack

Guide Agent on how to apply external knowledge:

## External Research Guidelines
- When suggesting new libraries, ensure compatibility with our existing stack
- Adapt external examples to match our coding standards and project structure  
- Verify that suggested solutions work with our publishing environment

Example request that leverages both: “Research the latest React 18 performance optimization techniques and implement them following the component patterns defined in replit.md” This approach gives you solutions that are both current and tailored to your specific project needs.

Limitations

File size and content limits

While there’s no strict character limit for replit.md, extremely large files may not be fully processed. Keep your replit.md focused and concise for best results.

Root directory requirement

replit.md must exist in your project’s root directory. Agent won’t automatically detect files in subdirectories.

Context scope

replit.md provides context for Agent conversations but doesn’t automatically apply to other AI tools.

Next steps

Ready to customize Agent for your project?

Start simple: Create a basic replit.md with your key preferences
Iterate and improve: Update your replit.md as you work with Agent
Share patterns: Use successful replit.md configurations across similar projects
Monitor effectiveness: Pay attention to how well Agent follows your guidelines and adjust accordingly

Learn more about working with Agent or explore other AI tools. Enterprise: Pre-configure replit.md in custom templates to give every builder in your organization a consistent starting context. For design-specific guidance, see setting up a design system.

​How replit.md works

​Setting up replit.md

​Automatic generation

​Manual creation

​Regenerating replit.md

​Best practices

​Be specific and clear

​Use examples

​Define communication preferences

​Specify project context

​Example replit.md configurations

​Web application project

​Data analysis project

​API development project

​Advanced usage

​Dynamic project guidance

​Integration with external tools

​Combining web search with replit.md

​Best practices for web search integration

​Limitations

​File size and content limits

​Root directory requirement

​Context scope

​Next steps

How `replit.md` works

Setting up `replit.md`

Automatic generation

Manual creation

Regenerating `replit.md`

Best practices

Be specific and clear

Use examples

Define communication preferences

Specify project context

Example `replit.md` configurations

Web application project

Data analysis project

API development project

Advanced usage

Dynamic project guidance

Integration with external tools

Combining web search with `replit.md`

Best practices for web search integration

Limitations

File size and content limits

Root directory requirement

Context scope

Next steps