MCP Server

Connect AI assistants (Cursor, Claude, VS Code, ChatGPT, Windsurf) to APIpie documentation using the Model Context Protocol (MCP).

What This MCP Server Provides (Today)

APIpie exposes a read-only MCP server that allows AI tools to:

• Discover documentation pages
• Fetch page content

This MCP server is documentation-only. It does not provide live platform APIs, routing execution, provider metadata, pricing, or write access.

MCP Endpoint

Use this MCP server URL:

https://apipie.ai/mcp

Available Tools

list-pages

Lists all available documentation pages (title, path, description).

get-page

Retrieves full markdown for a specific page.

What This MCP Server Does NOT Do

To avoid confusion for agent systems:

• No model routing or execution
• No provider/model inventory
• No pricing/quotas
• No account access
• No mutations (create/update/delete)
• No prompts (yet)

If it’s not listed under Available Tools, it does not exist.

Connect an AI Client

ChatGPT (Custom Connector)

1. Open ChatGPT Settings
2. Go to Connectors
3. Enable Developer mode (if required)
4. Create a New Connector
   • Name: APIpie Docs
   • MCP Server URL: https://apipie.ai/mcp
   • Authentication: None
5. Save

You can then enable the connector in a conversation and ask:

• “List APIpie docs pages”
• “Fetch the page for /routing/overview and summarize it”

Claude Code

claude mcp add --transport http apipie-docs https://apipie.ai/mcp

Claude Desktop

1. Open Claude Desktop → Settings → Developer
2. Click Edit Config
3. Add:

{
  "mcpServers": {
    "apipie-docs": {
      "command": "npx",
      "args": ["mcp-remote", "https://apipie.ai/mcp"]
    }
  }
}

4. Restart Claude Desktop

Cursor

Quick Install

If your environment supports MCP deep links, add the server in Cursor via Settings → Tools & MCP using:

• Name: apipie-docs
• Type: http
• URL: https://apipie.ai/mcp

Manual Setup

Create or edit:

.cursor/mcp.json

{
  "mcpServers": {
    "apipie-docs": {
      "type": "http",
      "url": "https://apipie.ai/mcp"
    }
  }
}

Visual Studio Code

Create or edit:

.vscode/mcp.json

{
  "servers": {
    "apipie-docs": {
      "type": "http",
      "url": "https://apipie.ai/mcp"
    }
  }
}

Ensure you have GitHub Copilot + Copilot Chat installed if you want Copilot to use it.

Windsurf

1. Open Windsurf → Settings → Cascade
2. Manage MCPs → View raw config
3. Add:

{
  "mcpServers": {
    "apipie-docs": {
      "type": "http",
      "url": "https://apipie.ai/mcp"
    }
  }
}

Zed

Add to your Zed settings JSON:

{
  "context_servers": {
    "apipie-docs": {
      "source": "custom",
      "command": "npx",
      "args": ["mcp-remote", "https://apipie.ai/mcp"],
      "env": {}
    }
  }
}

Validate It Works

After connecting, ask your assistant:

• “Call list-pages and show me the first 10 results.”
• “Call get-page for the routing overview page and summarize it.”

If those work, the MCP connection is correct.

Usage Examples (Docs-Only)

Once configured, AI assistants can reliably do:

• “List all APIpie documentation pages”
• “Get the authentication page and extract required headers”
• “Find the docs page describing routing fallbacks”
• “Retrieve the webhook docs and summarize event payloads”

Anything involving live routing, models, or account state requires APIpie’s API (not MCP).