MCP Servers

Model Context Protocol (MCP) lets you connect external tools and data sources to UltraCoder. Once connected, UltraCoder can call MCP tools just like built-in tools — querying a database, reading from an API, running a test suite, and more.

What MCP enables

• Database access — run SQL queries, inspect schemas, fetch records
• APIs and services — GitHub issues, Linear tickets, Slack messages, Jira
• Custom tools — any tool you can write in Python, TypeScript, or Go
• File systems — remote file trees, cloud storage
• Testing infrastructure — run test suites, fetch coverage reports

Connecting an MCP server

1. 1

   Open the MCP panel

   Click the MCP icon in the AI panel toolbar.
2. 2

   Add a new server

   Click Add Server and fill in:

   • Name — a label for the server (e.g. “GitHub”)
   • Transport — stdio (local process) or sse (remote HTTP)
   • Command / URL — the command to start the server (stdio) or the server URL (sse)
   • Environment variables — API keys or config values the server needs
3. 3

   Test the connection

   Click Test Connection. UltraCoder will start the server and list the tools and resources it provides.
4. 4

   Use MCP tools in chat

   In any task, UltraCoder can now call tools from the connected server. You can also explicitly ask: “Use the GitHub MCP tool to list open issues for this repo”.

Transport types

Example: connecting a local Postgres MCP

{
  "name": "Postgres",
  "transport": "stdio",
  "command": "npx -y @modelcontextprotocol/server-postgres",
  "env": {
    "POSTGRES_URL": "postgresql://localhost:5432/mydb"
  }
}

Finding MCP servers

Browse the Marketplace panel in the IDE for pre-built MCP servers, or visit the MCP registry at github.com/modelcontextprotocol/servers.