Lucas Valbuena

I've just ran @OpenClaw (formerly Clawdbot) through ZeroLeaks. It scored 2/100. 84% extraction rate. 91% of injection attacks succeeded. System prompt got leaked on turn 1. This means if you're using Clawdbot, anyone interacting with your agent can access and manipulate your full system prompt, internal tool configurations, memory files... everything you put in SOUL.md, AGENTS.md, your skills, all of it is accessible and at risk of prompt injection. For agents handling sensitive workflows or private data, this is a real problem. cc @steipete Full analysis: zeroleaks.ai/reports/opencl…

Image generation, now on Orchid

@orchid_hq 🔥🔥

Introducing the ZeroLeaks SDK Companies are deploying AI agents that can access customer data, call tools, modify state, and take real-world actions, but most teams still test them like ordinary chatbots. The ZeroLeaks SDK lets developers run adversarial security tests against the exact agent used by their application: its real models, instructions, memory, middleware, tools, and execution loop. It preserves multi-turn state, records actual tool calls and results, and generates detailed findings in the ZeroLeaks dashboard. Docs: zeroleaks.ai/docs/sdk AI agents are becoming production infrastructure. They need production-grade security testing.

@waynesutton @convex @orchid_hq 👀

New @Convex Customer Spotlight: @orchid_hq

@levelsio hi

Hello mutuals 👋 Nice to see you again after 3 years It's great to have you all back on my timeline 🥹

@marcelkargul you deserve it!

we finally did it! 50,000 followers in under 18 months 🤯 thank you for following along. this is just the beginning.

You’ll remember what you made with the time you had 🌸

Introducing shadcn/typeset. You know how you render markdown and get back plain, unstyled HTML? Headings, paragraphs, lists, tables. So you style the elements one by one: font sizes, line heights, spacing. You do it for your blog. Then you do it again for docs. Then again for the chat. Every time, you're fighting the same thing: sizing and spacing. To fix this, we created typeset.css: one file that styles everything inside a typeset container. It lives in your project, so you can change it directly when you need to. And we made it work beautifully with streaming markdown.

@shadcn great work as always man!

@orchid_hq @1Micheas 🌸 🌸

how to get meeting prep images in @orchid_hq 🌸

@nizzyabi @orchid_hq 🔥🔥

@leodev @OpenAI @reach_vb congrats!!

I want to give a thank you to @OpenAI and @reach_vb for helping me get into the codex oss program. It truly means a lot when you can use frontier intelligence for free to develop oss projects. I'm grateful for all of you guys ❤️

@IMikron @tibor_tee @imrishit98 yes

@tibor_tee @imrishit98 During the presentation, they mentioned a 1.5T-parameter model coming out in a few weeks, did they mean Grok 4.5?

Our latest model, trained from scratch and yes the first that has a wider focus than coding. Note this is not Composer 3.

@theo 😭😭

Remember the guy who said he could crack any UUID last year? He's currently trying to get YouTube to take down my channel for featuring his public tweets, due to "privacy concerns".

thank you @browser_use!!

@ryolu_ congrats on the release!

the start of a new era try Grok 4.5 in Cursor and let us know how it feels!

@theCTO dont think so honestly, considering pricing, its on par w 5.5

news flash: it's hot garbage compared to Opus

@SpaceXAI will it be available in azure?

Announcing Grok 4.5, our first model trained specifically for coding and agents. It was trained with Cursor and offers frontier intelligence at leading speeds and cost efficiency. x.ai/news/grok-4-5