MCP Server Setup

Connect your AI agent to Premex Support using the Model Context Protocol (MCP). Your agent gets tools to manage products and conversations, documentation resources, and guided prompts that walk you through setup — so you never have to leave your editor.

Quick Start

1. Add the MCP server

Claude Code

bash

claude mcp add --transport http premex-support https://support.premex.se/mcp

Or add to your project's .mcp.json manually:

json

{
  "mcpServers": {
    "premex-support": {
      "type": "url",
      "url": "https://support.premex.se/mcp"
    }
  }
}

Cursor / Windsurf — add the URL https://support.premex.se/mcp in your editor's MCP server settings.

2. Authenticate

When you first use the MCP server, your agent will open a browser window for you to sign in with GitHub. This is a standard OAuth flow — you authorize once and the token is stored automatically.

No manual token copying required.

3. Run the setup prompt

Ask your agent to use the setup-support prompt:

Let's get started with Support by Premex

The prompt guides your agent through the full onboarding:

Check installations — discovers your connected GitHub App installs and Slack workspaces
Pick a backend — helps you choose GitHub Issues, Slack threads, or your own SPI backend
Create a product — configures everything with the right adapter settings
Embed the widget — reads the widget docs and gives you the embed code for your site

You can also run the other prompts directly:

embed-widget — get the embed code for an existing product
setup-slack — connect a Slack channel as your support backend

Available Tools

Product Management

Tool	Description
`list_products`	List all products for your tenant
`create_product`	Create or update a product (GhCare, SlackCare, or SPI)
`delete_product`	Delete a product configuration

Installations

Tool	Description
`list_installations`	List GitHub App and Slack workspace connections
`list_slack_channels`	List public channels in a connected Slack workspace

Conversations

Tool	Description
`list_conversations`	List conversations for a customer
`get_conversation`	Get conversation detail with messages
`create_conversation`	Start a new support conversation
`post_message`	Post a message to a conversation
`close_conversation`	Resolve or close a conversation

Resources

The MCP server exposes product documentation as resources so your AI agent can read the guides it needs on demand:

Resource	Content
`docs://premex-support/widget`	Widget embedding guide (script tag, iframe, form page)
`docs://premex-support/slack`	Slack integration setup
`docs://premex-support/spi`	Support SPI overview
`docs://premex-support/spi/endpoints`	SPI endpoint reference
`docs://premex-support/api`	REST API reference
`docs://premex-support/android`	Android SDK documentation

Endpoint

POST https://support.premex.se/mcp

The server implements the MCP Streamable HTTP transport in stateless mode with OAuth 2.1 authorization. Authentication is handled automatically by your MCP client.