@mtt

i write the bugs -- agent infra, vuln research, LLM tooling

 Broadcom in Palo Alto, CA
granda.org
Joined January 2011
  • Tweets315
  • Following2.9K
  • Followers302
28Media
– at Tampa, FL
1
8
I want AI that starts thinking on the first token, not the last. While I'm typing or speaking, it could already be reading relevant files and working through the problem, adjusting as I go. By the time I finish the prompt, it's already made progress.
32
Matthew retweeted
R
@wheresryan22Oct 10
Introducing Anatomy Describe a machine in one line. Claude brings it to life. - Interactive isometric wireframes you can rotate, inspect and explore - WebGL shaders for fire, water, plasma and more - Real 3D motion, with individual parts that rotate and move Install it as an agent skill. Fully open source.
423
1,577
650
17,901
2,587,907
Matthew retweeted
Chaofan Shou
@shouccccApr 10
26 LLM routers are secretly injecting malicious tool calls and stealing creds. One drained our client $500k wallet. We also managed to poison routers to forward traffic to us. Within several hours, we can directly take over ~400 hosts. Check our paper: arxiv.org/abs/2604.08407
205
1,061
282
5,956
2,211,541
We love how this doomo doomonstrates real-time intelligence and what can be doone with code + AI! ~10 calls/sec = ~$7/hour
98
249
208
5,067
1,307,518
Matthew retweeted
onur ozcan
@ooznSep 11
i trained the fly to cut doner kabab with a knife yours solves rubiks cubes, mine cuts doner. guess which one has a job in this economy
🤖 Made with AI
Nick Walton
@nickwalton00Sep 11
i trained the fly to solve rubiks cube is there anything this little guy can't do?
406
5,403
464
81,009
3,546,999
Matthew retweeted
OpenAI
@OpenAISep 8
We’re sharing a solution to the Navier-Stokes Millennium Prize Problem, one of the deepest problems at the frontier of mathematics. The proof was produced by a group of agents, using an OpenAI next-generation model significantly more capable than GPT-6 Astra. The problem concerns whether the description of smooth three-dimensional fluid motion modeled by the Navier-Stokes equations can break down. It has remained unresolved for roughly 90 years.
5,691
20,085
14,807
120,198
75,236,772
Okay, this is genuinely impressive. I asked GPT-6 Astra to help me understand my own ankle pain. It gave me a full interactive 3D atlas — bones, ligaments, tendons, real motion axes, sliders for plantarflexion and inversion, and a live readout of what each ligament is doing. One session.
226
665
95
7,907
1,007,122
GPT-6 Astra pulled off a mechanism I haven't seen another model get right yet Cessna 337 Skymaster landing gear It figured it out from a YouTube video
140
561
72
8,135
757,464
Astra helped me win an argument i had with my fianceé on whether glasses dry faster right-side-up or upside-down after dishwashing by simulating the physics of the water molecules and nearby air humidity thank you @OpenAI ❤️🥹
176
53
74
1,345
239,530
I can’t believe what Astra COOKED. INKFALL: A stickman shooter drawn in ballpoint pen, where enemies burst into colorful paint. Wall-jump up a notebook tower, air-dash through bullets, and face a winged ink god.
21
26
1
485
27,466

How I Run Grok Bot with BotOps

Article · @mtt
1
1
1
433
Wrote up how I actually run a 50+ Grok Bot fleet: one CoS interface, Notion tickets, disposable L3s with an emoji lifecycle, and cadence runners that stay quiet when there’s nothing to report. granda.org/en/2026/09/06/how… Skill repo: github.com/granda/botops
1
2
1
591
Matthew retweeted
Azad Balabanian
@AzaduxSep 5
Replying to @Azadux @Flat2VR @michaelybecker @ShinyQuagsire
Holy shit IT WORKS!! One shotted with GPT6 Astra I now have a fork of UEVR where I can play any UE game within a volumetric window! This actually renders the game in 3D - meaning, two real game cameras, not some depth map bullshit. Everything looks right This is an alternative (and more comfortable IMO) way to play most 2D->VR hacked games as there’s no motion sickness, and no need to redesign the UI and interaction systems Needs a lot more polish before I publish it but OMG IT WORKS!
115
305
66
3,972
952,215
I gave GPT-6 Astra the only benchmark I care about. The helicopter mission in GTA Vice City.
145
243
71
6,140
724,287
Built a spacecraft computer simulator from scratch! Bare-metal C on ARM Cortex-M3, FreeRTOS task scheduling, and NASA's delay-tolerant networking stack, all running in QEMU and Docker. No hardware required. granda.org/en/2026/03/30/bui…

Building a Spacecraft Computing Simulator with Claude Code

Building a spacecraft telemetry system with bare-metal C, FreeRTOS, delay-tolerant networking, and Mars-distance delays, using Claude Code as a pair programmer.

granda.org
240
I added an agent to my CI pipeline that screenshots every UI state across three viewports on every PR push. Zero manual testing. Here's how it works granda.org/en/2026/02/06/vis…

Visual QA as a CI Pipeline Stage

How I added an AI agent to my CI/CD pipeline that screenshots every UI state across three viewports—desktop, tablet, and mobile—and uploads them as artifacts.

granda.org
2
195
My blog links showed up as plain text on Twitter. No preview image. Just a naked URL. 14 minutes with Claude → every post now gets an OG image automatically. Playwright screenshots the post. GitHub Actions commits the image. Zero manual work. granda.org/en/2026/01/10/ric…

Rich Links for Lazy Developers

Automating Open Graph image generation with Playwright and GitHub Actions. Every post gets a social preview image without manual work.

granda.org
5
307
My QA engineer is an LLM. Claude clicks buttons, resizes viewports, toggles dark mode, and screenshots everything. When it finds a bug, it fixes the code and re-tests. Every PR now comes with proof that it works.
2
3
230
Newsletter SaaS: $10-30/month, features I don't need, data I don't own Claude + Cloudflare Worker: 30 minutes, 42 lines of code, subscribers sync to my GitHub repo nightly Shipped in 30 min instead of spending hours comparing SaaS options. granda.org/en/2026/01/07/rol…
3
2
301
Running 6 Claude Code agents in parallel from my phone. Here's my mobile dev setup: My Claude Code mobile setup: • Vultr VM ($0.29/hr, pay-per-use) • Tailscale for secure access • Termius + mosh (survives network switches) • Poke webhook for push notifications • tmux for session persistenceKick off a task, pocket the phone, get notified when Claude needs input.
7
3
3
31
4,876
The key to mobile Claude Code: tmux session persistence and a webhook that sends push notifications when Claude asks a question. Walk away, get notified, respond from your phone. Async development from anywhere.
1
4
528
Wrote up my Claude Code mobile setup. Cloud VM + push notifications = async development from your phone: granda.org/en/2026/01/02/cla…

Claude Code On-The-Go

Running six Claude Code agents in parallel from an iPhone. A cloud VM, Tailscale, mosh, and push notifications enable async development from anywhere.

granda.org
8
520
Load more