#security

In my last post, I mentioned that I was “investigating the possibility to use Tailscale to simply block the access from the “outside world” and make it more secure”. Well, it is now implemented and it works. Basically, my blog system can listen on two different ports: the regular one where you’re navigating, and the one inside a Tailscale network.

If I specify exclusiveDashboard: true on my Tailscale section of the configuration, the dashboard will only be added to the Tailscale router. This way, the dashboard won’t be available at hacdias.com, but only inside the Tailscale network. I’ve tried it and it works. However, it is not how it is not enabled on the live version yet!

I also mentioned that I was trying to “inject” and admin bar while logged in. I can say that it is working via some response writer wrapping magic. I have a few more plans to make Eagle more self-contained, but we’re not there yet!

Have a fantastic day!

Weirdly enough, the name Tailscale caught my eye and I decided to investigate what it is. I checked their website and it seems that they use Wireguard to do the kind of configuration I explained some time ago. I’m curious: besides the easyness of use, what other advantages do you see on using Tailscale over a manually configured Wireguard network?

A few days ago I saw a blog post by GitHub announcing their support for security keys in SSH operations. After a bit of search, I discovered that OpenSSL 8.2, released last year, lets us use our FIDO2 devices (such as Yubikeys) to “unlock” our SSH private keys. This way, we don’t need the cumbersome GPG <-> SSH relation.

I tried it, and it was nice. I don’t know if I’m going to start using it right away because not every place uses OpenSSL 8.2+ yet, but soon, very soon.

Some links:

• https://xosc.org/u2fandssh.html
• https://buttondown.email/cryptography-dispatches/archive/cryptography-dispatches-openssh-82-just-works/

I enjoy showing the webmention and the context (reply to what? repost of what? like of what?) because, as you know, the content on the Internet is ephemeral and if I don’t store it, I don’t have assurance that it will remain available. And that’s the main reason why I show the webmentions.

In any case, it’s not the webmentions that worry me, but the post contexts that I show on replies, likes and reposts… Need to decide on that: either remove the pictures, or store them.

This is actually the next thing I want to “fix”. Other problem of twitter is that, if the user changes the profile picture, the previous one gets unavailable. For webmentions, I’m using the images provided by webmention.io, but for replies, likes and reposts context, I’m using the original source and I don’t like it.

I already ponderated to remove the images at all from the contexts, but I don’t think it would look as friendly as it does now. What do you think?