dt-app-dashboards

This skill teaches an agent to work with Dynatrace dashboards end to end: creating, modifying, querying, and analyzing the dashboard JSON documents (stored in the Dynatrace Document Store) that contain tiles, grid layouts, DQL queries, variables, and visualizations. It solves the problem of authoring correct, valid dashboards programmatically without breaking the strict structural rules Dynatrace enforces, such as matching tile IDs to layout IDs and using the 24-column grid.

It documents the dashboard JSON schema (content version, variables, tiles of type markdown or data, and layouts with x/y/w/h positioning) and a mandatory seven-step create/update workflow: define purpose and load domain skills, explore available fields, plan structure, design and validate every DQL query with `dtctl query`, download existing dashboard JSON before updating, construct or modify the JSON, and deploy with `dtctl apply`. Strong guardrails are emphasized: never invent DQL (load domain skills first), validate all queries before adding them, avoid time-range filters in tile queries since the UI picker handles that, and always download the current server state before editing so user UI edits are not silently overwritten. It also catalogs visualization types by data shape (time-series, categorical, single-value/gauge, tabular, distribution, maps, matrix) and their required field types, plus a variables reference covering query, csv, and text variable types with single- and multi-select filter patterns. Reference files cover create/update, tiles, variables, and analyzing.

It targets Dynatrace observability engineers, SREs, and platform teams who build or audit dashboards, and it is Apache-2.0 licensed. All operations go through the official dtctl CLI against the user's own tenant.