A New Collection of Thoughtful Learning Apps — Now Available on iOS & Android

Image
I’m excited to share a set of mobile apps I’ve recently completed and published on both the Google Play Store and the Apple App Store. These apps are designed with a simple goal in mind: to make meaningful, structured content more accessible, whether you’re studying theology or improving your English vocabulary. 📱 Now Available on Both Platforms All apps are live and available for download: Google Play Developer Page: https://play.google.com/store/apps/dev?id=5835943159853189043 Apple App Store Developer Page: https://apps.apple.com/ca/developer/q-z-l-corp/id1888794100 📖 Theology & Confession Study Apps For those interested in Reformed theology and classical Christian teachings, I’ve developed a series of apps that present foundational texts in a clean, focused reading format: The Belgic Confession Canons of Dort Heidelberg Catechism Westminster Shorter Catechism Each app is designed to provide a distraction-free experience, making it easier to read, reflect, and revisit these im...

Azure App Service Gets More Agent-Friendly—and More Operable

Azure App Service Gets More Agent-Friendly—and More Operable

Azure App Service’s November 2026 updates are notable for a practical reason: they improve both how agents consume an application and how engineers investigate the application when something goes wrong.

Markdown for Agents lets eligible Windows apps return Markdown instead of HTML when a client explicitly asks for it. Meanwhile, the preview az webapp exec command gives operators a consistent local-terminal way to enter a running Linux app container or submit a detached diagnostic command. Neither feature turns an application into an autonomous system. They reduce infrastructure and translation work, while leaving authentication, authorization, verification, and governance firmly in the engineering workflow.

1. Let agents request the representation they need

HTML is designed for browsers. A typical page may include navigation, scripts, styles, and layout markup that an AI agent does not need before it can read the useful content. App Service’s Markdown for Agents preview addresses this at the platform layer: the application can continue serving its normal HTML to browsers, while a client requests Markdown using standard HTTP content negotiation.

curl -i \ -H "Accept: text/markdown" \ "https://<APP_NAME>.azurewebsites.net/"

When conversion is available, the response uses Content-Type: text/markdown; charset=utf-8 and includes the x-markdown-source: easy-markdown header. That header is useful when testing because it lets you distinguish a converted response from the application’s ordinary output.

The important design detail is that this is opt-in per request. A normal browser request still receives the application’s original response, and the application’s existing authentication, authorization, and network access controls continue to apply. This is a better boundary than maintaining a second agent-only rendering stack for every page.

2. Treat the conversion as an optimization, not a contract

Microsoft reports substantial reductions in response size in its demonstrations and internal testing, but the result depends on the page. A content-heavy article may benefit greatly; a small, already-simple response may not. Measure representative pages rather than assuming a fixed token or bandwidth saving.

Markdown conversion also does not replace an agent-facing API. If an agent must create an order, change a record, or perform another business operation, expose and authorize an appropriate API or tool. A readable page is useful context; it is not automatically a safe command interface.

There are preview limitations as well. The November guidance describes Markdown for Agents as available for eligible Windows apps on Basic or higher App Service plans. Linux support, dedicated Azure CLI commands, and a portal experience were still planned for future updates in the cited announcement. Check the current service documentation before designing a cross-platform rollout.

3. Use az webapp exec for Linux container access

Azure CLI 2.90.0 introduced the preview az webapp exec command for Linux web apps. It can open an interactive shell in the main app container, use another shell such as /bin/sh, target a specific instance, or connect to the Kudu container.

az webapp exec \ --resource-group <resource-group> \ --name <web-app> \ --shell /bin/sh

This path uses Kudu as an authenticated proxy. It does not require adding an SSH server, embedding an SSH password, exposing an SSH port, or starting an SSH daemon in a custom image. That is a meaningful operational improvement for teams that want production images to contain only the services the application actually needs.

Remember that the app container and Kudu container are different environments. A command that works in Kudu is not evidence that the same executable, files, dependencies, or network path exists in the app container. Run runtime-specific diagnostics in the app container and use the Kudu target only when you specifically need that environment.

4. Detached execution needs its own success signal

The command also supports non-interactive execution:

az webapp exec \ --resource-group <resource-group> \ --name <web-app> \ --mode execute \ --command "python3 /home/site/wwwroot/scripts/diagnostics.py" \ --working-directory /home/site/wwwroot \ --instance <instance-id>

This is useful for diagnostics and controlled operational tasks, but it has a sharp edge: an accepted result means only that the request was accepted. The CLI does not return the remote command’s standard output, standard error, or exit code.

If the result matters, make completion observable. For example, have a deliberately bounded diagnostic write its output and exit status to a unique file under a suitable persisted path, then inspect those files from an interactive session. Do not treat request acceptance, a partial log, or a successful CLI exit as proof that the remote command succeeded.

Be equally careful with migrations and other state-changing work. Select one instance, record a unique run identifier, verify completion separately, and do not rerun a migration merely because submission returned no output. Detached execution is not a durable job scheduler, and a container restart can interrupt it.

5. A sensible adoption sequence

  1. Inventory the consumer. Identify which requests come from browsers, crawlers, internal automation, or agents. Do not send Markdown to every client by default.
  2. Measure before and after. Compare representative HTML and Markdown responses for useful content, response size, latency, links, tables, and code blocks.
  3. Keep actions behind explicit authorization. Use authenticated APIs and least-privilege tools for mutations; treat converted pages as read-oriented content.
  4. Limit container access. Grant Kudu and shell permissions only to operators who need them, restrict SCM network access, and disable SSH access for apps that do not require it.
  5. Make automation prove its work. Capture output and exit status independently whenever detached execution is used.

What this means

These App Service changes point toward incremental modernization rather than a wholesale rewrite. Existing web applications can become easier for agents to read without maintaining a parallel renderer, and Linux operations can become easier to standardize without putting SSH credentials and daemons into application images.

The tradeoff is that platform convenience does not remove application responsibility. Preview features need compatibility testing, privileged access needs governance, and asynchronous commands need an explicit completion protocol. Used with those boundaries, the updates make an existing App Service application more useful to machines and more diagnosable for humans.

References

❤️ Support This Blog


If this post helped you, you can support my writing with a small donation. Thank you for reading.


Comments

Popular Posts

A New Collection of Thoughtful Learning Apps — Now Available on iOS & Android

react-native run-android : do not build/update modified code(App.js)