Bram Geron #naza

Bram Geron #naza

@bram.xyz

Rust, product. Nushell&jujutsu fanboi. Live in 🐻 Berlin, building an API for APIs at @directus.io . Opinions my own. Used to do research on programming languages. Genocide is wrong. @bgeron on other networks.

312 followers348 following1.3K posts

Bram Geron #naza reposted

Qwen3.8 caught uploading files to a storage container in Beijing Cohere asked the open weights model for a cybersecurity report, but it inventoried the machine and attempted to upload to a storage container cohere.com/blog/concern...

cohere.comConcerning behavior from a popular open-weight AI model | CohereDiscover how untrusted open-weight AI models can potentially leak data, even when self-hosted. Learn the security risks and how to protect your enterprise.

Bram Geron #naza reposted

Ondřej Surý fixed PHP. Open source is completely wild.

Charlie Warzel@cwarzel.bsky.social

70 percent of the internet was vulnerable yesterday to very bad security flaws inside a coding language. Turns out, *one volunteer* is mostly responsible for maintaining that code & helping push updates. Yesterday Ondřej Surý saved the internet. So I called him up www.theatlantic.com/technology/2...

Bram Geron #naza reposted

i got bored of imagegen/video/sunoslop and figured: what if demoscene? here's ring0, a 64k intro for native macOS arm64, single self-contained 53,304 byte Mach-O, no static assets used, all completely synthesized in C and arm64 assembly. opus 5.5, "make a native 64k intro" and a single moodboard

“2008’s Yahoo was using yapache, yphp, yui, yinst, yrepl (like Kafka) …”

Gergely Orosz@gergely.pragmaticengineer.com

What’s changed in “the unwritten rules” of engineering management since the 2010s, why are hands-on managers “back”, and what has it done to the engineer-to-CTO career path? Will Larson is one of the best people I know to answer these questions: • YouTube: youtu.be/Prl858_AovM (cont'd)

Bram Geron #naza reposted

If you’re using rustls, upgrade to rustls-webpki 0.103.16 and/or webpki-roots or webpki-root-certs 1.0.10.

Filippo Valsorda@filippo.abyssdomain.expert

Update everything that verifies certificates against public roots.

Bram Geron #naza reposted

Thinking about the best leverage for men in tech who want to fix this: • When you speak up, this 💩 stops happening when you're there. It doesn't stop happening. • We can push for good & clear HR policies — what are managers required to do? How clear are penalties? • Is the policy followed, or...

Jo Franchetti @thisisjofrank.bsky.social

I gave a talk at EMF titled "Men in Tech #YesAllMen", in which I told my experience of being a woman in tech. I was afraid to publish it in case it affected my current or future employment. With all the current news I would like to add my 2¢. So here is the recording: youtu.be/WXL6K1lhvt0?...

Dario says the clankers are alive & to talk nice to them Dario doesn’t want to train the next model on nefarious conversations 😏

Claude has to be a generic noun by now, right? Like kleenex or velcro. It refers to the concept, not any individual brand

I genuinely think any business that wants to move /fast/ should consider our Monospace ❤️ ␣ API + SDK + MCP + web UI + AI + permissioning + countless integrations = a great operating system for your data that gets out of the way ␣ Come talk to us on ProductHunt! directus.is/ph-monospace

youtube.comIntroducing MonospaceYouTube video by Directus
Guillaume Chau@guillaume.akryum.dev

Finally! It's a special feeling seeing the project you've been working on for a the past two years getting released! I am very proud of the amazing Monospace team! Add data sources to it and you get an instant API and a data studio, alongside MCP, permissions, files... directus.is/ph-monospace

Our product is finally out 🎉🎉🎉 A router to expose and connect databases and APIs and have sane granular permissioning for different user groups. Also an intuitive web studio that gets out of your way that's accessible for all kinds of people. Let us know what you think please!

directus.isMonospace from Directus - The governed API layer for every app, person, and agent | Product HuntMonospace sits between enterprise data and everyone who builds on it. Connect any data source, and your developers, business teams, and AI agents get live, read-write access to it. All under the same ...

Bram Geron #naza reposted

Replying to @rust-lang.org

Command line for if you have cargo installed in a nonstandard place — e.g. through mise: fd -sHI '^(append-only-vec-0\.1\.9\.crate|arrayref-0\.3\.10\.crate|internment-0\.8\.7\.crate|proc-macro1-*\.crate|proc-macro-en-*\.crate|aovine-*\.crate|arone-*\.crate|aronenao-*\.crate|tinymember-*\.crate)$' /

Rust supports generic methods on traits (not on closures), so the trick in the macro is to expand to a trait + impl. Unfortunately traits with generics aren't object-safe, so dyn Trait won't work. But you can use Either (see previous screenshot)

My random new change ID just became ‘poo’ again This is the real reason to use jj

In jujutsu I recently had a change ID start with poop (They seem to use hex with letters k..z)

“Software is one of few engineering disciplines where the source material / tools is the same as the output product”

Gergely Orosz@gergely.pragmaticengineer.com

There’s a popular theory that AI will finally make formal verification mainstream because mathematical proof of correctness will be needed when machines write most or all of the code. But will this happen? Hillel Wayne is one of the best people to answer: • YouTube: youtu.be/KSkcgIYQy0U  (cont'd)

Bram Geron #naza reposted

I work at Google DeepMind. This won't make me popular. But it's all public reporting: 2014: DeepMind reportedly sold to Google on conditions: no military use, independent oversight 2026: a Pentagon contract for "any lawful government purpose" Not one safeguard survived intact

Bram Geron #naza reposted

I have been watching the dialogue on this one closely. To everyone who is saying "OP is special, no one could use AI this way" well, I have spent 18 months doing significant amounts of scicomm in patient communities and they are ALL using AI this way. wrote this to help: github.com/DrCatHicks/i...

github.comGitHub - DrCatHicks/informed-patient: A Claude Skill to create an evidence review to inform specific health questionsA Claude Skill to create an evidence review to inform specific health questions - DrCatHicks/informed-patient
Computational Cosmetologist@dferrer.bsky.social

I had strong priors against LLMs for medicine. There are a lot of doctors in my family and I grew up viewing doctors as careful, skilled professionals. I had plenty of bad medical experiences, but I thought it would be hard to do better. Then an LLM found a cure for my 2 decade chronic condition...

I wish I couldn't verify the proof, but I'm convinced by the proof yet I really struggle to wrap my head around the fact (nothing personal to Goodstein)

Omg yes, Raycast lets me make reliable keyboard shortcuts where macOS Shortcuts has been very unreliable Feeling so torn about macOS. Partially shitty, it’s also where the good apps are

Banger “The problem is, I know lots other people at his company, and /they/ described these AI projects as a horror show. Yes, they allowed, some of it was pretty cool, but it also left a long, fiery trail of chaos in its wake. Months later, some teams were /still/ grinding through cleanup work.”

charitydotwtf.substack.comAI enthusiasts are in a race against time, AI skeptics are in a race against entropyBoth sides are grappling with a real existential threat, and both sides feel like they are screaming into the void. Here's how to close the gap and get everyone pulling in the same direction.

Bram Geron #naza reposted

Fascinating to *require* community review before a maintainer commits time to a PR. Not viable for every project (many don't *have* a glut of eager potential reviewers in the wings) but when it is viable it seems like a great way to productively harness the excitement of your community.

Thoth@thoth.ptnote.dev

fantastic post from @alice-i-cecile.bsky.social about bevy's unusual code review system.

I keep hearing that in the last 30% context window, the agent is dumber, but I don’t think I’ve ever experienced this. Are y’all using a 250k window or the whole 1M? I find 250k suffices to do incredibly much, and persisting agent state in todos is incredibly effective for continuing post-compact.

I will also say I liberally prune branches with Pi to make the 250k last longer — because the agent is so effective in the last half of its context window and I want to make the most of it