Composable slots​

The headline change is that slots are composable: Header, Sidebar, Content, and Details can live anywhere in your component tree—inside an <Outlet />, a <Suspense> boundary, or any wrapper—instead of having to be direct children. That's what makes it work cleanly with routers, where the layout frame and the routed content can own different slots. Beyond that, panel state is preserved across collapse and expand, the details panel can render side-by-side or as an overlay, and panels are responsive with their own breakpoints.

Migrating from Page​

Most of the migration is mechanical, so you just need to rename the slots and a few props. A few APIs changed along the way (expand/collapse controls are always on, content always stays mounted) and the default breakpoints are different, so review any breakpoint overrides after you switch.

Start with the PageLayout component and the migration guide.

New for app builders​

• Custom page titles—set a descriptive browser-tab title with the usePageTitle hook, so users can tell multiple open tabs apart. See Page titles.
• Lazy loading—a deeper guide on code splitting, route- and view-based lazy loading, when not to split, and how to parallelize a chunk download with its Grail query. See Understand lazy loading.

From the Strato design system​

• PageLayout—the new, more flexible replacement for Page, with composable slots and a split-or-overlay details panel. See the PageLayout component and its migration guide.
• Accessibility foundation—a new design-foundation page covering what you need to build inclusive apps — WCAG 2.2 AA, and what Strato handles for you. See Accessibility.
• TimeframeSelector—from @dynatrace/strato-components v3.1+, it shows users their recently used timeframes automatically — no code changes required. See TimeframeSelector.
• XYChart—a low-level, composable option for charts the ready-made components can't build — heatmaps, scatterplots, and more. See XYChart.

From the community​

One thread we highlighted: a user asked whether they could set a custom page title in a Dynatrace app. It was on the roadmap then, and it's now shipped, with the usePageTitle hook above. Your questions genuinely feed what we prioritize.

Next session​

The next App Development Office Hours is Tuesday, September 8, 16:00 CEST / 10:00 EDT. We'll post the sign-up link on Dynatrace's LinkedIn closer to the date, so follow along there to register.

Got a topic you'd like us to cover? Drop it in the Developer Q&A Forum.

Lazy loading: ship less upfront​

By default, an app bundles every page and component into one download, even the ones the user never opens. Lazy loading splits that bundle into chunks that load on demand, using React's lazy() and <Suspense>. The App Toolkit now supports this out of the box.

The highest-impact, lowest-risk place to start is the route level: load only the page the user navigates to. From there, you can defer in-page views that sit behind an interaction (tabs, drawers, modals), and even fire a Grail query while the browser is still downloading the component that will render it. The guide also covers where not to split: anything above the fold, or smaller than ~20–30 KB, usually isn't worth a separate chunk.

Stale-while-revalidate: reuse what you've fetched​

Grail queries can take seconds, and much of the data an app shows such as host lists, topology, or configuration, doesn't change second to second. Stale-while-revalidate (SWR) renders cached data instantly, then quietly refreshes it in the background and swaps in the new result if anything changed. No spinner, no blank screen.

You mostly get this for free: useDql from @dynatrace-sdk/react-hooks is built on TanStack Query and ships with SWR support. The guide walks through setting staleTime per data type, using placeholderData to avoid a flash of loading state on filter changes, and the key distinction between isLoading (first load) and isFetching (background refresh).

Why both matter​

The two techniques map neatly onto Core Web Vitals: lazy loading reduces the JavaScript the browser parses before first render, which helps LCP; while SWR removes loading states from repeat navigation, which helps INP.

Who it's for​

R&D engineers who build or contribute to Dynatrace Apps — whether you've just started with Claude Code or already use it daily and want a deeper understanding of the underlying concepts.

What you'll learn​

• How to get the most out of Claude Code—context, memory, and project conventions
• How MCP works, and how to connect Claude to live Dynatrace data
• How to write, share, and evaluate Skills for your team
• What AI Agents are and how they differ from Skills
• How to use Spec-Driven Development to agree on what to build before writing code, and let AI do more of the execution

Format​

The workshop is virtual and runs a full day (09:00–16:00 CET, with breaks). The hands-on exercises run real Claude Code sessions, so budget roughly $10–20 USD in usage for the day.

There's a bit of setup involved: Node.js, Claude Code, the demo app, and the Dynatrace MCP servers. We recommend you check the prerequisites and get them done before the session.

Sign up​

Sessions are scheduled and open for enrollment. Find the dates, the full agenda, and the prerequisites on the AI-First App Development Workshop page on Advance Galaxy.

Stuck on the setup, or have questions before the day? Reach out in #team-platform-success.

You add a small usePageTitle hook to your project and call it from each routed page. The app shell composes the rest of the tab title (app name, tenant, "Dynatrace") around the part you set. One thing to watch: only set a title for views with a persistent URL — updating it for transient overlays or modals confuses the browser's back/forward history.

The Page titles guide has the full walkthrough, including the hook itself and how to set titles for widgets.

This one closed a loop with the community — it started as a question on the Developer Q&A forum. If there's something you're missing, that's a good place to ask.

• Conversation starters add an Explain with Dynatrace Assist action to any app.
• Built-in prompts push that same idea into the dashboards and notebooks you ship.
• Contributing DQL examples helps train the NL2DQL model that turns natural language into DQL.

Conversation starters​

Conversation starters connect any app on the platform to Dynatrace Intelligence, so users can get AI-powered context without leaving what they're doing. Think of an Explain with Dynatrace Assist button, pre-loaded with the right question and context. It works via an intent to the Dynatrace Assist app, with a prompt and optional supplementary or instruction context.

A couple of things to know before you build:

• Conversation starters currently run in non-agentic mode, and because LLM output isn't deterministic, you add your use cases to the regression test suite via GitOps.
• Always check availability first with listAvailableSkills—not every tenant or user has Dynatrace Assist enabled—and only render the trigger when the conversation skill is present.

Built-in prompts for ready-made dashboards​

This takes the same idea as conversation starters a step further. Instead of a button, you integrate Dynatrace Intelligence directly into the dashboards and notebooks that ship with your app. The setup is lightweight: add a prompt field to your DQL result set and users get a Dynatrace Assist trigger right there in the data. Add execute=true to auto-send it, and pass hidden context to make the response more relevant to what the user is looking at.

Contribute DQL query examples​

Rather than looking examples up at query time, the NL2DQL model is trained on hundreds of thousands of real DQL queries. You can help by contributing well-documented examples to improve the model, and increase the chance it gets the queries you care about right. We're especially short on RUM-on-Grail and FinOps examples.

Questions​

For anything Dynatrace Intelligence–related, reach out in #help-dti-agentic-ai.

What's new is the new home for app developer news on Dynatrace Developer. It's a single, searchable, filterable feed of news worth knowing, without all the channel-hopping.

What you'll find here​

This isn't a changelog—you'll still find those in Release notes.

In What's new, you can expect a steady mix of updates about new features and ideas, interesting events and tools for your work, plus lots of other topics that matter to app creators.

Today, you see the MVP. More features are coming soon. Use the feedback button at the bottom of the page to tell us what you want more of (or less of).