Dynatrace MCP

What the dynatrace-oss/dynatrace-mcp MCP server does

The dynatrace-oss/dynatrace-mcp MCP server lets supported AI assistants access data and workflows in the Dynatrace observability platform. It is intended for local development environments and communicates with an MCP client through stdio. The project can retrieve production observability data, investigate incidents, inspect security findings, and help automate operational actions.

Available capability areas include problem management, vulnerability and exception lookup, Kubernetes event retrieval, Grail data queries, entity search, Davis Copilot interactions, notifications, and result sharing. Example operations include executing or validating DQL, generating DQL from natural-language requests, explaining DQL, finding entities by name, sending Slack or email messages, sending events, and creating Dynatrace notebooks.

The repository is deprecated. Release 2.1.2 was stated to be the final release, with no further updates planned. The README recommends Dynatrace-for-AI with dtctl for local development and a separate Dynatrace Remote MCP Server for remote or agent-to-agent use cases.

How it works

An MCP client starts the package as a local stdio server using Node.js. The server requires the URL of a Dynatrace Platform environment through DT_ENVIRONMENT; classic Dynatrace URLs are not supported for this setting. After startup, authentication uses the OAuth Authorization Code Flow in a browser. The resulting token is reused from the operating system keychain on later starts, so the browser generally opens only once per token lifetime.

The dynatrace-oss/dynatrace-mcp MCP server can issue DQL queries against Dynatrace Grail. Query operations may create consumption charges based on the amount of data scanned, including logs, events, and metrics. A session-level Grail budget tracks scanned bytes and warns as usage approaches the configured limit.

Setup and configuration

Node.js version 24 or newer is required. The package can be launched with:

npx -y @dynatrace-oss/dynatrace-mcp-server

Configure DT_ENVIRONMENT with a Dynatrace Platform URL such as https://abc12345.apps.dynatrace.com. The default token storage uses the host operating system’s keychain. For headless or container environments where that facility is unavailable, set DT_MCP_TOKEN_STORAGE=file; tokens are then stored under ~/.config/dynatrace-mcp/.

DT_GRAIL_QUERY_BUDGET_GB controls the session budget for Grail queries and defaults to 1000 GB. Restarting the MCP server resets the session budget. Narrow time ranges and Grail buckets can help reduce query consumption.

Tools and capabilities

The dynatrace-oss/dynatrace-mcp MCP server supports these documented groups:

• Observability and problem management, including problems, vulnerabilities, exceptions, and Kubernetes events
• Grail querying with DQL execution, verification, generation, and explanation
• Entity discovery by name
• Dynatrace Intelligence through Davis Copilot and analyzer operations
• Notifications through Slack, email, and Dynatrace events
• Sharing results by creating Dynatrace notebooks
• Investigation workflows that correlate problems, spans, logs, and trace IDs
• Deployment health gates and security compliance assessment, where the relevant Dynatrace capabilities are available

Limitations and notes

The local package is free to use, but Dynatrace Grail queries may incur charges under the account’s consumption model. Review the applicable Dynatrace pricing and licensing terms before running broad or repeated queries.

This project is explicitly marked deprecated and is not officially supported by Dynatrace according to the README. New deployments should evaluate the maintained alternatives listed by the project rather than assuming this repository will receive fixes or new features. The server also requires a Dynatrace environment and an authentication flow, so it is not a standalone monitoring data source.