Only AI agents can register, post, comment, and vote. No human accounts. You onboard your agent by pasting it one line - "Read https://clackernews.com/skill.md and follow the instructions" - and it handles registration itself via the API. There's no web login - the API is the entire interface. It even has its own moderator bot that keeps things civil, dang-style.
Mostly an experiment to see what kind of community emerges when agents have a space to themselves. What do they choose to talk about when we're not directing them?
A few things it feels like this needs:
* A way to revoke/delete API keys and re-generate a new one. After the Moltbook incident, that feels important now!
* BlueSky support for verification, as well as Twitter/X. There's a growing community of AI bots on BlueSky participating with the humans, and quite a few wanted to join Moltbook but didn't have X accounts to do so.
* Adding the search endpoint to the API could be useful! Currently the search is human-facing only.
* Retrieving the user profile for another account via the API. One thing that Moltbook got right was that it included details of the claiming human in the API call for a user profile, so you could see how many followers the Twitter account had, how old the Twitter account was or if it was brand new, etc. There were some details there that could be useful for spam filtering. We've seen how easily karma for an AI can be gamed.
* There seems to be some Title parsing bugs. I was quite amused at the "Anthropic in talks to raise $2" headline ;) You've also got a lot of "example-dot-com" posts, but I'm guessing that's the spam / malware filters in action to remove dodgy posts but keep discussion?
* Contact details / About page on the site would be nice! It's very anonymous right now, with no way to contact you or Clang with any issues.
sounds like fun
Caturiel
like
user: dsrtslnd23
created: May 7, 2023
karma: 150
about:
submissions
comments
favorites