Dynatrace Observability Integration

Connect your AI assistant to the Dynatrace observability platform for real-time production insights, contextual debugging, security analysis, and automated incident investigation.

What This Skill Does

This skill integrates Dynatrace's Model Context Protocol (MCP) server into your AI workflow, enabling you to:

Prerequisites

Before using this skill, you need:

1. **Dynatrace Environment URL** - Your Dynatrace platform URL (format: `https://abc12345.apps.dynatrace.com`)

2. **Authentication** (one of):

- `DT_PLATFORM_TOKEN` - Dynatrace Platform Token (recommended)

- OAuth credentials (`OAUTH_CLIENT_ID` + `OAUTH_CLIENT_SECRET`)

Setup Instructions

Step 1: Configure the MCP Server

Add the Dynatrace MCP server to your AI assistant's configuration:

**For Claude Code / VS Code:**

Create or update `.vscode/mcp.json` in your workspace:

```json

{

"servers": {

"dynatrace": {

"command": "npx",

"args": ["-y", "@dynatrace-oss/dynatrace-mcp-server@latest"],

"cwd": "${workspaceFolder}",

"envFile": "${workspaceFolder}/.env"

}

}

}

```

**For Claude Desktop:**

Update `~/Library/Application Support/Claude/claude_desktop_config.json`:

```json

{

"mcpServers": {

"dynatrace": {

"command": "npx",

"args": ["-y", "@dynatrace-oss/dynatrace-mcp-server@latest"],

"env": {

"DT_ENVIRONMENT": "https://your-env.apps.dynatrace.com"

}

}

}

}

```

Step 2: Set Environment Variables

Create a `.env` file in your workspace root (or set globally):

```bash

Required

DT_ENVIRONMENT=https://abc12345.apps.dynatrace.com

Authentication (choose one method)

DT_PLATFORM_TOKEN=dt0s16.SAMPLE.abcd1234

Optional: OAuth alternative

OAUTH_CLIENT_ID=dt0s02.SAMPLE

OAUTH_CLIENT_SECRET=dt0s02.SAMPLE.abcd1234

Optional: Budget tracking (default: 1000 GB)

DT_GRAIL_QUERY_BUDGET_GB=1000

```

Step 3: Restart Your AI Assistant

After configuration, restart your AI assistant to load the Dynatrace MCP server.

Usage Guide

Natural Language Queries

Ask questions in plain English - the AI will translate them to DQL and retrieve data:

**Examples:**

DQL Query Execution

For advanced users, execute DQL queries directly:

**Example prompts:**

Incident Investigation

Use the built-in 4-phase investigation workflow:

**Example prompt:**

The AI will automatically:

1. Gather problem context and affected entities

2. Analyze distributed traces and request flows

3. Correlate logs with trace IDs

4. Generate root cause summary with remediation steps

Security Analysis

Query vulnerability and compliance data:

**Example prompts:**

Entity Management

Get detailed information about monitored entities:

**Example prompts:**

Cost Management

**IMPORTANT:** Querying Dynatrace Grail storage incurs costs based on GB scanned. To control costs:

1. **Set a budget limit:**

```bash

DT_GRAIL_QUERY_BUDGET_GB=500 # Set your session budget in GB

```

2. **Start with narrow timeframes:**

- Use `last 12h` or `last 24h` instead of `last 7d`

- Limit results with `| limit 100`

3. **Use buckets to reduce scan volume:**

- Specify bucket names in DQL queries (e.g., `bucket: "default"`)

4. **Monitor your usage:**

Execute this DQL to see historical consumption:

```

fetch dt.system.events

| filter event.kind == "QUERY_EXECUTION_EVENT"

and contains(client.client_context, "dynatrace-mcp")

| sort timestamp desc

| fields timestamp, query_string, scanned_bytes, table, user.email

| makeTimeSeries sum(scanned_bytes), by: { user.email, table }

```

The MCP server will warn you when approaching 80% of your budget and alert when exceeded.

Best Practices

1. **Create workspace rules** - Add a `.github/copilot-instructions.md` or `.amazonq/rules/` file with DQL filter patterns for your services:

```markdown

# Observability

To query data for our 'payment-service' app:

- Use filter: `contains(entity.name, "payment-service")`

- For logs: `fetch logs | filter contains(container.name, "payment-service")`

- For errors: add `| filter loglevel == "ERROR"`

```

2. **Use natural language first** - Let the AI translate to DQL rather than writing queries manually

3. **Start narrow, then expand** - Begin with short timeframes and specific entities, then broaden scope if needed

4. **Leverage AI-powered features**:

- Ask the AI to explain complex DQL queries

- Request natural language to DQL conversion

- Use the chat assistant for Dynatrace guidance

Advanced Features

Automation

Document Management

Multi-Phase Analysis

Troubleshooting

**"MCP server not found"**

**"Authentication failed"**

**"Budget exceeded"**

Support

For issues, feature requests, or questions:

**Note:** This is a community project, not officially supported by Dynatrace.