Skip to main content

​
What are MCP Servers?

Think of MCP servers as plugins that give the AI access to:
  • 🗄️ Your databases (Supabase, PostgreSQL, etc.)
  • 🔍 Web search and documentation
  • 🌐 External APIs
  • 🤖 External tools (even Figma, Blender, etc.)
This allows the AI to not just write code, but also query your database, search documentation, or interact with your services.

​
Built-in MCP Servers

Teta includes several pre-configured MCP servers:

​
1. Supabase MCP Server

Purpose: Interact with your Supabase database Capabilities:
  • Query database tables
  • Read schema information
  • Execute SQL queries (read-only)
  • Browse database structure
Setup: Automatically configured when you provide SUPABASE_ACCESS_TOKEN

​
2. Dart MCP Server

Purpose: Flutter/Dart development assistance Capabilities:
  • Dart package documentation
  • API references
  • Best practices
  • Code examples
Setup: Enabled by default for Flutter projects

​
Adding Custom MCP Servers

You can add your own MCP servers through Teta Secrets.

​
How to Add

  1. Inside a project, go to Settings > MCPs
  2. Click “Add MCP”
  3. Add a name
  4. Add the json configuration, for example:
{
  "command": "npx",
  "args": ["-y", "@modelcontextprotocol/server-github"],
  "env": {
    "GITHUB_PERSONAL_ACCESS_TOKEN": "$GITHUB_TOKEN"
  }
}
The json content will be saved as a secret to protect any sensitive info (API keys, tokens, etc.) When an MCP server is added to a project, its content cannot be viewed again.

​
Configuration Example

Secret Name: server-github Secret Value: 
{
  "command": "npx",
  "args": ["-y", "@modelcontextprotocol/server-github"],
  "env": {
    "GITHUB_PERSONAL_ACCESS_TOKEN": "<GITHUB_TOKEN>"
  }
}

​
PostgreSQL Server

Package: @modelcontextprotocol/server-postgres Capabilities:
  • Query database
  • Read schema
  • Execute SQL
Configuration: 
{
  "command": "npx",
  "args": ["-y", "@modelcontextprotocol/server-postgres"],
  "env": {
    "POSTGRES_CONNECTION_STRING": "$DATABASE_URL"
  }
}

​
Puppeteer Server

Package: @modelcontextprotocol/server-puppeteer Capabilities:
  • Web scraping
  • Screenshot capture
  • Browser automation
Configuration: 
{
  "command": "npx",
  "args": ["-y", "@modelcontextprotocol/server-puppeteer"]
}

​
Brave Search Server

Package: @modelcontextprotocol/server-brave-search Capabilities:
  • Web search
  • Real-time information
  • Research assistance
Configuration: 
{
  "command": "npx",
  "args": ["-y", "@modelcontextprotocol/server-brave-search"],
  "env": {
    "BRAVE_API_KEY": "$BRAVE_KEY"
  }
}

​
Using Multiple MCP Servers

You can add as many MCP servers as you need. Each one adds different capabilities to the AI. Just add each as a separate secret in Settings > MCPs.

​
Security Best Practices

​
1. Use Read-Only Access

When possible, use read-only tokens: 
{
  "args": [
    "-y",
    "@supabase/mcp-server-supabase",
    "--read-only",
    "--project-ref=${PROJECT_REF}"
  ]
}

​
2. Limit Scope

Only grant necessary permissions:
  • For GitHub: Use personal access tokens with minimal scopes
  • For databases: Create read-only users
  • For APIs: Use restricted API keys

​
Finding More MCP Servers

You can find ready-to-use MCP servers at: Most MCP servers come with installation instructions on their GitHub/npm pages.

​
Conclusion

MCP servers let you connect the AI to your existing tools and services, making it more powerful and context-aware.
Need help? Contact support@teta.so