MCP Server

Connect AI assistants to Send247's delivery platform using the Model Context Protocol.

AI Agents
ChatGPT
Claude
Copilot

The Send247 Delivery MCP is a Model Context Protocol (MCP) server that allows AI assistants to quote and book UK deliveries through a structured, safe, and auditable interface.

🌐 MCP Endpoint

https://mcp.send247.uk/api/v1/mcp

What This Does

The MCP server exposes delivery functionality as tools that an AI assistant can safely invoke:

Real-time Quotes

Get delivery options and prices between UK addresses instantly.

AI Booking

Create deliveries using a previously generated quote with user confirmation.

Safe by Design

Enforces confirmation before irreversible actions. Fully auditable.

Available Tools

get_delivery_quote
Read

Returns delivery options and prices for a shipment. Use when a user wants to know delivery costs or compare services before booking.

Stateful (caches quotes for booking)
Non-destructive
Uses external delivery providers
json
// Tool: get_delivery_quote
// Returns delivery options and prices for a shipment
 
// Example prompt to AI assistant:
"Get a delivery quote from London to Manchester"
 
// The AI will call the tool and return pricing like:
{
"quote_id": "qt_abc123",
"pickup": "London, UK",
"dropoff": "Manchester, UK",
"options": [
{
"service": "same_day",
"price": 2499,
"currency": "gbp",
"estimated_delivery": "Today by 6pm"
},
{
"service": "express",
"price": 3999,
"currency": "gbp",
"estimated_delivery": "Within 2 hours"
}
],
"expires_at": "2024-01-15T16:00:00Z"
}
create_delivery
Write

Creates a delivery booking using a previously issued quote. Must only be invoked after explicit user confirmation.

Writes data (creates bookings)
Destructive (irreversible)
Requires user confirmation
json
// Tool: create_delivery
// Creates a delivery booking using a previously issued quote
 
// Example prompt to AI assistant:
"Yes, book the same-day delivery option"
 
// The AI will confirm before calling the tool:
"I'll book the same-day delivery for £24.99.
This will create an irreversible booking. Confirm?"
 
// After user confirms, creates the delivery:
{
"delivery_id": "del_xyz789",
"status": "confirmed",
"tracking_url": "https://track.send247.uk/del_xyz789",
"pickup": {
"address": "London, UK",
"scheduled": "Today 2pm-4pm"
},
"dropoff": {
"address": "Manchester, UK",
"estimated": "Today by 6pm"
}
}

Client Configuration

This MCP server is publicly accessible and can be connected to by any MCP-compatible client. No client-side API keys are required as authentication is handled at the gateway level.

ChatGPT / OpenAI

mcp-config.json
{
"mcpServers": {
"send247-delivery": {
"url": "https://mcp.send247.uk/api/v1/mcp"
}
}
}

Once connected, enable the app in a conversation and prompt normally (e.g., "Get a delivery quote from London to Manchester").

Claude (Anthropic)

Note: Claude Desktop & Claude Code run MCP servers locally and don't yet support direct remote MCP URLs. Use the official MCP remote bridge.

First, install MCP Remote:

bash
npm install -g @modelcontextprotocol/remote

Then add to your MCP config:

claude-mcp-config.json
{
"mcpServers": {
"send247-delivery": {
"command": "npx",
"args": [
"@modelcontextprotocol/remote",
"https://mcp.send247.uk/api/v1/mcp"
]
}
}
}

Claude will automatically discover available tools via tools/list.

GitHub Copilot (VS Code)

vscode-mcp-config.json
{
"mcpServers": {
"send247-delivery": {
"url": "https://mcp.send247.uk/api/v1/mcp"
}
}
}

When using Copilot Chat, tools will appear when delivery-related prompts are detected.

MCP Manifest (Optional)

For platforms or registries that support MCP manifests, you may include:

mcp.json
{
"name": "Send247 Delivery",
"description": "UK delivery quotes and booking via the Model Context Protocol",
"version": "1.0.0",
"endpoint": "https://mcp.send247.uk/api/v1/mcp",
"capabilities": {
"tools": true
}
}

Safety & Design Guarantees

Confirmation Required

No deliveries created without explicit user confirmation

Quote Expiry

Quotes expire automatically to prevent stale pricing

Schema Validation

Tool schemas are strictly validated

Server-side State

Ensures reliability across LLM sessions

Auditable Actions

All write actions are logged and auditable

UK Only

Designed for end users in the United Kingdom

Intended Use

This MCP server is designed for:

  • AI assistants offering delivery services
  • Agentic workflows that require logistics actions
  • End users in the United Kingdom

It is not a general logistics SDK and does not expose raw provider APIs.

📧 Support

For integration questions or issues, contact us at [email protected]