Header image

It's full of stars

Where documentation meets reality


Update Astro with the help of AI

By Tobias Hofmann November 17, 2026 Posted in REST

Reading time: 3 min read


Astro versions and their release date (as given by Google).

Version Publish date
Astro 4.0 5. December 2023
Astro 5.0 3. December 2024
Astro 6.0 10. March 2026
Astro 7.0 22. June 2026

When I started to use Astro for my personal blog, I opted to use Astro 4.x. I started the first draft versions of my blog in early 2025. At that time, Astro v4 was already outdated, as v5 had been available for a few months. The problem I faced was that the information I found on the internet was mostly about v4. Content regarding v5 was not widely available. Also, the AI tooling available at that time struggled to deliver the same level of quality results for v5 as they did for v4.

Plan

My blog went live on Astro v4 in January 2026. In 2026 version 6 and 7 of Astro were released. In November 2026 that left me three versions behind. I wanted to update to Astro 7. I looked at the official upgrade guides 🔗. Guides are available to get me from 4 to 7. It looks like a direct upgrade from 4 to 7 is not possible (or not documented). But: going from 4 to 5 to 6 to 7 is documented. I decided to let Claude do the work for me.

The prompt was simple: Plan an upgrade of my blog on Astro 4 to latest version 7. Look at the official documentation on how to do this. Validate after each version upgrade that the site still works.

Claude gave me a plan: upgrade version by version to Astro 7. Check all dependencies. Update node and Astro plugins. It found out several dependencies between versions, what to upgrade first and how to validate that the static build of the site still works. Testing also involved running the site before the upgrade, picking some posts as a reference and validating that those still work after each version upgrade. Of course, all work was done on a feature branch.

Execute

The plan looked good, so I gave my OK. Claude started, and after a few minutes the first upgrade was done. A few minutes later Astro 6 was installed. After a total of maybe 30 minutes, Astro 7 was up and running. The most time-consuming part was the site build process, as it took a few minutes to create the complete site. Some findings made it into the final report, for instance regarding performance.

Done

This was fast. I guess finding out all those dependencies between plugins, Astro, Node.js etc. would have taken me more than a few minutes. Not to mention reading through the documentation.

I had time left, so I asked Claude to clean up the code. Old, commented-out and unused code was removed. New features like image handling were added. Some bugs regarding mobile access and article handling were also fixed, some of them simply because I asked Claude to do a code review. Every change was validated via Playwright, automatically. When something did not look good, I only had to take a screenshot, paste it and say: the heading is wrong.

Result

My personal blog is now using Astro 7. Several improvements are now implemented and live. Time needed: not even one hour. The live deployment, including the first build, took 7 minutes. That’s fast. And the result? The site works. During the migration, constant quality assurance was provided. I only had to ask for it.