AppDynamics MCP Server

A Model Context Protocol (MCP) server for integrating AppDynamics with Claude Code. This allows you to query AppDynamics data directly from Claude Code for debugging production issues, analyzing metrics, and investigating errors.

Features

• Query applications, nodes, tiers, and backends
• Fetch metrics and performance data
• Get health rule violations and alerts
• Retrieve transaction snapshots and error details
• Support for both Production and Non-Production environments

Prerequisites

• Node.js 18 or higher
• AppDynamics account with API access
• Claude Code (or any MCP-compatible client)

Installation

1. Install dependencies:

cd appdynamics-mcp-server
npm install

2. Build the TypeScript code:

npm run build

3. Configuration is already set up at ~/.appdynamics-mcp/config.json

Configuration

Your configuration file is located at ~/.appdynamics-mcp/config.json and contains credentials for both production and non-production environments.

Security Note: Never commit this file to version control. The config directory is in your home folder and should have restricted permissions.

Adding to Claude Code

Add this server to your Claude Code configuration file (~/Library/Application Support/Claude/claude_desktop_config.json on macOS):

{
  "mcpServers": {
    "appdynamics": {
      "command": "node",
      "args": [
        "/Users/vkrishnaraj/work/projects/appdynamicsclaude/appdynamics-mcp-server/dist/index.js"
      ]
    }
  }
}

After adding the configuration, restart Claude Code.

Available Tools

1. list_applications

List all applications in the specified environment.

Parameters:

• environment (required): "prod" or "nonprod"

Example:

Can you list all applications in production?

2. get_application

Get details for a specific application.

Parameters:

• environment (required): "prod" or "nonprod"
• applicationName (required): Name of the application

Example:

Get details for the "lftracingservice" application in production

3. get_metrics

Get metrics for an application.

Parameters:

• environment (required): "prod" or "nonprod"
• applicationName (required): Name of the application
• metricPath (required): Metric path (e.g., "Overall Application Performance|*")
• durationInMins (optional): Duration in minutes (default: 60)

Example:

Get performance metrics for lftracingservice in production for the last 2 hours

4. get_business_transactions

Get business transactions for an application.

Parameters:

• environment (required): "prod" or "nonprod"
• applicationName (required): Name of the application

Example:

Show me business transactions for the agentservice application in prod

5. get_health_rule_violations

Get health rule violations for an application.

Parameters:

• environment (required): "prod" or "nonprod"
• applicationName (required): Name of the application
• durationInMins (optional): Duration in minutes (default: 60)

Example:

Are there any health rule violations for lftracingservice in production in the last hour?

6. get_snapshots

Get transaction snapshots for an application.

Parameters:

• environment (required): "prod" or "nonprod"
• applicationName (required): Name of the application
• durationInMins (optional): Duration in minutes (default: 60)

Example:

Get transaction snapshots for fraudsvc in production from the last 30 minutes

7. get_error_snapshots

Get error snapshots for an application.

Parameters:

• environment (required): "prod" or "nonprod"
• applicationName (required): Name of the application
• durationInMins (optional): Duration in minutes (default: 60)

Example:

Show me errors in the lftracingservice application in production

8. get_nodes

Get nodes for an application.

Parameters:

• environment (required): "prod" or "nonprod"
• applicationName (required): Name of the application

Example:

What nodes are running for production-compliance in prod?

9. get_tiers

Get tiers for an application.

Parameters:

• environment (required): "prod" or "nonprod"
• applicationName (required): Name of the application

Example:

Show me tiers for the lftracingservice application

10. get_backends

Get backends for an application.

Parameters:

• environment (required): "prod" or "nonprod"
• applicationName (required): Name of the application

Example:

What backends does the agentservice application connect to?

Usage Examples

Once configured in Claude Code, you can ask natural language questions like:

• "Show me all applications in production"
• "What errors occurred in lftracingservice in the last hour?"
• "Get health rule violations for fraudsvc in nonprod"
• "Show me slow transactions for production-compliance"
• "What's the response time for lftracingservice in the last 2 hours?"

Troubleshooting

Server not connecting

1. Check that the build completed successfully: npm run build
2. Verify the path in your Claude Code config matches your actual installation path
3. Check that config.json exists and has the correct credentials

Authentication errors

1. Verify credentials in ~/.appdynamics-mcp/config.json
2. Test credentials by making a direct API call:

curl -u "nettracer@nettracer:garaz19xw0qr" \
  "https://nettracer.saas.appdynamics.com:443/controller/rest/applications?output=JSON"

No data returned

1. Verify the application name is correct (case-sensitive)
2. Check the time range - increase durationInMins if needed
3. Ensure the application exists in the specified environment

Development

To make changes to the server:

1. Edit files in src/
2. Rebuild: npm run build
3. Restart Claude Code

For continuous development:

npm run watch

Security

• Configuration file contains sensitive credentials
• Keep ~/.appdynamics-mcp/config.json secure with appropriate file permissions
• Never commit credentials to version control
• Consider using environment variables or secret management for production use

License

MIT