I’ve been focused on making ClarkLab easier to set up, safer to update, and less annoying to fix when something breaks.

The biggest part of this was improving the update flow. After adding health checks, I worked on making updates more reliable by testing candidate containers before replacing the live API. This means ClarkLab can now check that the new version is actually healthy before switching over, instead of restarting and hoping it works. It should make updates more stble, especially when running it on your own server.

I also spent time improving agent updates. Agents can now be updated using release binaries, and I added more work around prebuilt agent binaries, release patches, and version handling. This should make agent updates cleaner and less dependent on building everything manually on the target machine.

Another big focus was setup. I added an api and Cloudflare setup wizard to make self-hosting easier. The goal is to guide people through the annoying parts like env setup, production configuration, Cloudflare tunnel setup, frontend variables, and install scripts. It is nowhere near perfect and used agentic assistance, but it is a big step toward making it easier to deploy without needing to manually configure every little option across multiple files.

I also improved the node setup flow. There is now a cancel setup button, so if you start setting up a node and mess something up, you are not just stuck with a half-created node. I also patched several small bugs such as the node install command.

For debugging, I updated agent logs to use the journal. This should make logs more useful and easier to understand when something goes wrong on a server. I thought this was important as i want this project to be usable for real infra. This feature is very much so in an early stage as i am nowhere near finished with it.

There were also a bunch of security and maintenance fixes. I patched agent authentication, blocked node modification for users who are not sysadmins, strengthened password reset handling, reduced the chance of sensitive data being exposed through frontend logs, and tightened GitHub repo parsing. I also updated workflow permissions and CI checks so changes are safer.

This devlog is again, not about new frontend features. It is mostly about making this project more stable, easier to deploy, and safer. The main thing I learned from the last ship is that setup still needs to be much easier, so I started moving toward a proper guided setup flow. I very much so want to keep improving that until someone can deploy ClarkLab without having to understand how it works completely first.