How We Built an MCP Server to Handle Your Uptime Monitoring#

What is the Model Context Protocol (MCP)?#

MCP Architecture Diagram

Why We Built an MCP Server for Uptime Monitoring#

1. Check monitoring status through conversation - Simply ask your AI assistant about your website's uptime
2. Analyze incidents naturally - Get insights about outages without navigating dashboards
3. Create new monitors on the fly - Set up monitoring for new endpoints through simple conversation
4. Receive contextual recommendations - Get AI-powered suggestions based on your monitoring data

Technical Deep Dive: How Our MCP Server Works#

1. Server Initialization: The server registers with the MCP protocol and declares its capabilities
2. Tool Registration: We define a set of functions (tools) that AI assistants can call
3. Authentication: Your API key is securely stored and used for all API calls
4. Request Handling: When an AI assistant invokes a tool, our server processes the request and calls the uptime-agent.io API
5. Response Formatting: Results are formatted and returned to the AI assistant in a way it can understand

• listMonitors - Retrieves all your active monitoring endpoints
• getMonitor - Gets detailed information about a specific monitor
• createMonitor - Sets up a new endpoint to monitor
• listIncidents - Views all detected downtime incidents
• getIncident - Gets detailed information about a specific incident
• listIncidentsByMonitor - Shows all incidents for a particular endpoint
• createAnonymousTracking - Creates public tracking without authentication

Setting Up the MCP Server#

Prerequisites#

• An active uptime-agent.io account
• Your uptime-agent.io API key (found in your account dashboard under Settings → API Keys)
• Node.js 18 or higher installed

Option 1: Quick Install via NPM (Recommended)#

1npx uptime-agent-mcp setup

• Install the MCP server
• Configure it for use with Claude Desktop
• Prompt you for your uptime-agent.io API key
• Set up all necessary configurations automatically

Option 2: Install via Smithery.ai#

1. Create an account at smithery.ai
2. Get your personal key from your Smithery account
3. Run the following command:

1npx -y @smithery/cli@latest install @AVIMBU/uptime_agent_mcp --client claude --key <personal_key>

Option 3: Manual Configuration in Claude Desktop#

1. Open Claude Desktop on your computer
2. Navigate to Settings (gear icon) → Advanced Settings
3. Find the "MCP Servers" section and click "Add New Server"
4. Enter the following configuration:

1{
2  "mcpServers": {
3    "uptime-agent": {
4      "command": "npx",
5      "args": [
6        "-y",
7        "uptime-agent-mcp"
8      ],
9      "env": {
10        "UPTIME_API_KEY": "your-api-key-here"
11      }
12    }
13  }
14}

1. Replace your-api-key-here with your actual uptime-agent.io API key
2. Save the configuration

Step 4: Test the Connection#

"Claude, can you list all my active uptime monitors?"

"I've connected to your uptime monitoring system. You currently have 5 active monitors:

1. Production API (https://api.example.com/health) - Status: UP
2. Marketing Website (https://example.com) - Status: UP
3. Customer Portal (https://portal.example.com) - Status: UP
4. Payment Gateway (https://payments.example.com/status) - Status: WARNING
5. Analytics Dashboard (https://analytics.example.com) - Status: DOWN

Would you like more details about any specific monitor or the current incidents?"

Using the MCP Server in Your Daily Workflow#

Checking Monitor Status#

"Give me a summary of all my monitoring endpoints."

Creating New Monitors#

"Please create a new monitor for our blog at https://blog.example.com. Check it every 5 minutes."

Analyzing Downtime#

"What incidents occurred last week and what was the average response time before resolution?"

Investigating Specific Issues#

"Tell me more about the incident on our payment gateway from yesterday."

Security Considerations#

1. API Key Security: Your API key is stored locally and only transmitted over HTTPS to authenticate with our API
2. Limited Permissions: The MCP server can only access endpoints authorized by your API key
3. No Data Storage: The server doesn't store any of your monitoring data locally
4. Secure Communications: All API calls use HTTPS encryption

Future Vision: The Autonomous Uptime Monitoring Agent#

Our Roadmap to Autonomous System Management#

1. Natural Language Analytics: Ask questions about your uptime patterns in plain English
2. Proactive Notifications: Have Claude alert you when incidents occur
3. Slack Integration: Connect your Slack channels for team notifications
4. Multi-Account Support: Manage multiple uptime-agent.io accounts from a single MCP server

The Agentic Web Approach to System Reliability#

• Automatic Deployment Integration: Direct communication with your Docker containers, Kubernetes clusters, and PaaS solutions
• Intelligent Log Analysis: Parsing through system logs to identify root causes without human intervention
• Autonomous Recovery Actions: Automatically restarting services, scaling resources, or implementing fixes when issues are detected
• Predictive Failure Prevention: Identifying patterns that precede outages and taking preventative action
• Continuous Learning: Improving its troubleshooting capabilities through each incident it resolves

Conclusion#