MCP Overview

Understanding Model Context Protocol and Dashtray

MCP Overview

Model Context Protocol (MCP) is a new standard that allows AI agents to safely interact with external tools and services.

What is MCP?

MCP is a standardized interface for AI models to:

• Request tools - Ask what tools are available
• Call tools - Execute specific functions
• Receive results - Get structured responses

Think of MCP as a standardized way for Claude, Cline, and other agents to “call” functions on external services.

How Does Dashtray Use MCP?

Dashtray provides an MCP server that exposes the send_notification tool. When you configure Dashtray with Claude or Cline:

1. Agent requests tools - “What can you do?”
2. Dashtray responds - “I can send notifications”
3. Agent calls tool - “Send a notification about this…”
4. Dashtray executes - Notification arrives on your phone
5. Agent continues - Can use the result in its workflow

Why MCP?

For Users

• 🔒 Secure - Tools are properly sandboxed
• 🛠️ Standardized - Works across different agents
• 🔄 Composable - Combine multiple MCP servers

For Developers

• 📝 Clear Interface - Standardized tool definitions
• 🚀 Easy Integration - Simple to add to agents
• 📚 Discoverable - Agents can inspect available tools

Dashtray as an MCP Server

Dashtray acts as an MCP server with:

Name: dashtray

Tools:

• send_notification - Send notifications to your device

Features:

• ✅ Real-time delivery
• ✅ Error handling
• ✅ Rate limiting

For AI Agents: llms.txt

Dashtray publishes an llms.txt file — a plain-text brief designed for AI assistants. It explains what Dashtray does, how to send a notification (MCP tool and raw HTTP webhook), plan limits, and links to every doc page in Markdown form.

You can paste that URL into any AI assistant’s context, or point an agent at it when you want it to learn Dashtray’s capabilities on its own:

Common Use Cases

1. Build Notifications

Get alerts when builds complete or fail:

Claude: "When the build finishes, send me a notification"
User: "Done! The build completed in 2m 45s"
Claude: sends notification via Dashtray

2. Task Reminders

Get reminded about important findings:

Claude: "I found a critical security issue. Let me notify you"
Claude: sends notification via Dashtray

3. Automation Status

Monitor long-running automation tasks:

Cline: "Deploying to production..."
Cline: sends notification when complete

4. Agent Confirmations

Get approval before taking important actions:

Agent: "Ready to deploy. Confirming via notification"
Agent: sends notification asking for approval

Architecture

┌─────────────────────┐
│  Claude Desktop     │
│  or Cline           │
└──────────┬──────────┘
           │
           │ MCP Protocol
           │
┌──────────▼──────────┐
│  Dashtray MCP       │
│  Server             │
└──────────┬──────────┘
           │
           │ REST API + Auth
           │
┌──────────▼──────────┐
│  Dashtray Backend   │
└──────────┬──────────┘
           │
           │ Push Notifications
           │
┌──────────▼──────────┐
│  Your Device        │
│  (iOS/Android/Web)  │
└─────────────────────┘

Setup Overview

1. Generate API Key - Create a key in Dashtray
2. Configure Agent - Add Dashtray MCP to Claude/Cline
3. Pass API Key - Provide the key as environment variable
4. Start using - Agents now have access to send_notification

Next Steps

• 🧠 Set up Claude Desktop
• 🤖 Set up Cline
• 📖 Tool documentation
• 💡 Examples

Learn More

• MCP Spec
• Claude Desktop Docs
• Dashtray API Reference