Live data from Hacker News

Moltworker: a self-hosted personal AI agent, minus the minis

blog.cloudflare.com

71–74 of 74 posts

Re: Moltworker: a self-hosted personal AI agent, minus the minis

#71
post #12

I have a bespoke local agent that I built over the last year, similar in facilities to Moltbot, but more deterministic code. Running it this kind of agent in the cloud certainly has upsides, but also: - All home/local integrations are gone. - Data needs to be stored in the cloud. No thanks.

This is exactly the issue. Even if you ignore the privacy concerns, the reason ClawdBot/Moltbot/OpenClaude got so popular is that everything was actually run locally. The early adopters where people on locked down corporate networks where almost everything they need to interact with is in the category of "a local printer" (possibly a networked one).

Cloudflare simply cannot access anything most users will want to access. If it's not run locally, it simply won't work for most users.

Piled on top is the obvious data privacy issue. Most notably the credential privacy, but also the non-credential privacy and data collection. Hard pass from me until there's a solution that covers all of these, including personal data privacy (and a "privacy policy" is no privacy at all).

Re: Moltworker: a self-hosted personal AI agent, minus the minis

#72

The prompt injection concerns are valid, but I think there's a more fundamental issue: agents are non-deterministic systems that fail in ways that are hard to predict or debug. Security is one failure mode. But "agent did something subtly wrong that didn't trigger any errors" is another. And unlike a hacked system where you notice something's off, a flaky agent just... occasionally does the wrong thing. Sometimes it…

How are these agents stress tested today? Are there tools that are typically being used for QA and/or security?

Re: Moltworker: a self-hosted personal AI agent, minus the minis

#73
post #48

Clawdbot/Moltbot looks to be a supply-chain attack waiting to happen, and I pity the poor soul who finds out when this ticking time bomb eventually detonates.

That’s what first came to my mind, the multiple integrations and cascaded connections probably will introduce multiple attack vectors. But, what’s the hype with motlbot anyway? I can just open any AI app and ask whatever, especially moltbot already uses the same AI vendors.

The point is that it has access to a TON of tools, permanent memory and can run "independently", or it's started by a background process to check if there's anything to do.

So you can tell it stuff like "I'm going to a concert March 3rd, it's outdoors so it might be cancelled due to weather, check the event's web page and tell me if there are any notifications". And then it'll just decide itself how to organise the work, setting notifications for itself to "wake up" to do something later, figuring out how to access the event page and read it.

There was one anecdote (of fan fiction, you can't really tell these days) where one user's Openclaw had pre-emptively messaged their partner that "I'm going to be working late today" because the bot saw the person had multiple work-related things open and a long todo-list still incomplete.

Re: Moltworker: a self-hosted personal AI agent, minus the minis

#74
post #12

I have a bespoke local agent that I built over the last year, similar in facilities to Moltbot, but more deterministic code. Running it this kind of agent in the cloud certainly has upsides, but also: - All home/local integrations are gone. - Data needs to be stored in the cloud. No thanks.

I've been thinking of a similar thing, I just need a local model with consistent tool calling performance.

Most of my crap could just be tools and a mid-level language model interpreting the results and deciding whether to act on them.

Post reply on HN