@dynatrace-oss/dynatrace-mcp-server

Exposes Dynatrace monitoring and observability APIs as standardized MCP resources, enabling LLM clients to query infrastructure metrics, application performance data, and logs through a unified protocol interface. Implements MCP resource discovery and schema advertisement, allowing clients to introspect available Dynatrace data sources without prior knowledge of the API structure.

Registers Dynatrace API operations as callable MCP tools with schema-based function signatures, enabling LLM clients to invoke monitoring queries, retrieve metrics, and fetch logs through structured function calls. Implements parameter validation and response marshalling to ensure type safety between LLM-generated function calls and Dynatrace API contracts.

Manages Dynatrace API token lifecycle and authentication headers for all outbound API requests, supporting environment variable configuration and secure credential passing. Implements request signing and token injection at the HTTP layer, ensuring all MCP tool calls and resource queries are properly authenticated against Dynatrace endpoints.

Executes parameterized queries against Dynatrace metric and log APIs, translating high-level query requests into properly formatted Dynatrace API calls with time range handling, filtering, and aggregation. Implements query result parsing and normalization to present data in consistent JSON structures regardless of underlying Dynatrace API response format.

Implements MCP resource listing and schema advertisement endpoints that allow clients to discover available Dynatrace data sources and their query parameters. Dynamically generates resource schemas based on Dynatrace API capabilities, enabling clients to understand available metrics, logs, and entities without hardcoded knowledge of Dynatrace structure.

Implements error handling for Dynatrace API failures including rate limiting, authentication errors, and malformed responses. Translates Dynatrace API error codes into MCP-compatible error responses with descriptive messages, enabling clients to understand and handle failures gracefully without exposing raw API error details.