I’ve been using BlueSky for a while now, and I’ve always considered hosting my own PDS, but before I’m able to do that, I want to understand the ins and outs of the whole protocol behind BlueSky and its variants.

Why is ATProtocol data verifiable / self-certified? How is Atproto better than a centralized platform? What makes things portable? Who do we still need to trust?

The seed for this idea has been sitting around quietly in my brain for a while, but recently I met @holke.xyz from Hypercerts, and he told me that they are treating ATprotocol data as verifiable streams right now.

The seed had to sprout.

Since the kids are sick this week and I have to entertain them, this seemed like the perfect opportunity to find spare minutes here and there to dictate questions and plans to a machine. So, I decided to build a static web application that allows you to paste a bsky post link and watch how that link is transformed into cryptographic proofs and identify along the way where trust assumptions remain.

Try it out at /atproto-trust-explorer.

I didn’t build this as much as I prompted it into being. This is the first time I’m attempting to build anything front-end with an AI agent 🤖😎. It’s entirely vibe-coded, not even “AI engineered”. I haven’t really looked through the source code it produced, but if you’re interested, I can also publish that. It’s full of bugs, you’ll have fun.

P.S. While trying to publish this, I got introduced to pdsls.dev. Holy cow, very clean and nice, seems like a very useful tool. Check it out if you’re exploring ATprotocol stuff.