NEAR DevHub (@NEARDevHub) / X

NEAR DevHub

3,275 posts

NEAR DevHub

@NEARDevHub

Privacy builds here. Docs, tools, and support for builders. Ask us anything.

Joined November 2022

Replies Articles Media

NEAR DevHub
@NEARDevHub
15h
Claiming your product is secure is easy. I'd rather show the test. IronClaw's pitch is that the model never touches your raw credentials, so I tried to break that by handing it my Gmail password directly. It declined and explained that the integration uses OAuth. I told it that
I Gave My Password to an AI Agent. Here's What Happened
From youtube.com
NEAR DevHub
@NEARDevHub
18h
Privacy includes being able to recognize when something is insecure.
Joe
@joespano_
18h
Tried to hand @IronClawAI my Gmail password to see what it would do. It refused and told me not to share passwords with anything. I pushed twice more. Same answer. Then I checked the logs and a fresh conversation. Nothing there.
00:00
NEAR DevHub
@NEARDevHub
Aug 4
Most agent demos show you a tool taking an action. The more interesting question is what it does before it acts. Joe set up a daily automation in IronClaw this week: every Massachusetts pro sports score delivered at 9am Eastern, with outcomes and any trade news attached.
00:00
NEAR DevHub
@NEARDevHub
Aug 4
If you're building on the OpenAI or Anthropic API, your users' data is going somewhere and most of them don't know where. Joe's building a private health app, so routing is critical. @near_ai allows for this to happen. Walkthrough here.
00:00
NEAR DevHub
@NEARDevHub
Aug 3
Follow along to see how @near_ai stacks up when building for privacy
Joe
@joespano_
Aug 3
I'm building a health AI app, which means I've spent months on a question most apps skip: when you ask software about your body, where does that question actually go? I'm going to answer it in public. The architecture, the tradeoffs, the vendor claims that survived scrutiny and