Researchers at @wiz_io put Gemini 4 Argon to the test against complex cyber threats.
In real-world testing, it found more than two times more vulnerabilities than baseline models — catching a critical risk that previous frontier models missed.
See how they did it ↓
Pieter van Dijk retweeted
N20092 (second SIA 777-9) completed 3h18m maiden flight 16 Oct and returned to Paine. RAT deployed for systems testing. Still targeting 2027 deliveries
Pieter van Dijk retweeted
This browser app teaches you Arduino circuit by circuit with 3D visuals and live code.
- without physical hardware
Perfect for beginners (and kids).
Pieter van Dijk retweeted
PHP powers 70% of the web.
Now Remote memory disclosure on any PHP application.
Today disclosed PHP2Shell remotely exploitable memory corruptions that enable pre-auth RCE on major CMSs and memory disclosure on essentially any PHP app.
- Two critical memory-corruption issues in PHP itself.
- (CVE-2026-109618/CVE-2026-109619)
- Unauthenticated RCE demonstrated on WordPress,
- Drupal, and Joomla.
Any PHP application can also leak process memory.
Patches released today upgrade as soon as possible.
- php2shell.com/#upgrade-php
Pieter van Dijk retweeted
🔴 Airbus' 2026 Cargo Global Market Forecast projects the worldwide fleet of dedicated freighter aircraft to grow 46% from 2,360 in 2026 to 3,460 aircraft by 2045. According to the forecast, there will be a requirement for 2,670 freighters.
Pieter van Dijk retweeted
Let’s go: DeepSeek just open-sourced DeepInfer, the inference engine behind its production services.
Sharing this makes open models efficient at scale is a big deal: model weights alone don’t make AI cheap to run. But *this* helps.
Also notable: it runs on Huawei chips, not just Nvidia.
🚀 Introducing DeepInfer, our in-house inference framework, now fully open source!
Powering DeepSeek-V4.1-Flash at scale with high throughput and low latency. We’re sharing our production stack to help the community lower compute costs, reduce energy use, and make AI more affordable.
🔗 github.com/deepseek-ai/DeepI…
1/4
Pieter van Dijk retweeted
The client timed out. Your API may have already saved the order.
If the client sends the request again, the server needs to know whether this is the same order or a new one.
You can give each intended order a key, then keep it for every retry. The idea is to store the result so a later retry can get it back.
Two requests can arrive at the same time, though. A cache check followed by a write leaves a gap where both can run. You have to claim the key in a single step so that only one request can win.
Save the order and its result in the same database transaction. If the app crashes before it replies, a retry can read the saved result. A payment provider also needs a way to detect repeats.
My ASPNET Core guide shows the response flow and explains the cache example's race window: milanjovanovic.tech/blog/imp…
Pieter van Dijk retweeted
Where should your AI agent run untrusted code?
An AI agent needs a workspace to run commands, install tools, and save files while completing a task.
It may also generate code or download it from external sources.
We should treat it as untrusted code and run it in an isolated workspace with limited access.
This is where Kubernetes Agent Sandbox comes in.
In our recent newsletter, we covered the following
- Why AI agents need an isolated workspace to run untrusted code
- How Kubernetes Agent Sandbox manages these workspaces
- How gVisor and Kata Containers isolate agent workloads
- Hands-on: Deploy a Kubernetes troubleshooting agent and run untrusted code
and more..
𝗥𝗲𝗮𝗱 𝗶𝘁 𝗵𝗲𝗿𝗲: newsletter.devopscube.com/p/…
Note: Kubernetes Agent Sandbox API is currently in Beta
♻️ PS: Repost and share it with the DevOps community.
#devops #mlops #aiops
Pieter van Dijk retweeted
A masterclass on how to build an unkillable SaaS in the AI era (from someone who bootstrapped and sold his company for $92,000,000): xitter.cf/lior_pozin/status/2069…
Pieter van Dijk retweeted
If you want to integrate Skillspector into your agent or project, just give this skill to Claude Code
github.com/NVIDIA/SkillSpect…
It's worth scanning your skills, especially if you've been installing random ones you find on X
Claude Code Templates now scans every component with NVIDIA SkillSpector
You can now see security scan results across Skills, Subagents, Commands, Settings, Hooks, MCPs, and Mods
SkillSpector analyzes components for potential security risks, including:
- Prompt injection and malicious instructions
- Data exfiltration and credential leaks
- Privilege escalation and tool misuse
-Supply chain vulnerabilities
- System prompt leakage and memory poisoning
A curated, security scanned, community maintained marketplace is far more valuable than millions of skills with unknown origins or unreviewed content
Security scanning isn’t a guarantee, but it’s an important step toward a safer agent ecosystem
aitmpl.com
Pieter van Dijk retweeted
AWS finally made centralized logging simple 🎉
Everybody says "you need a central logging account", but setting it up was quite a hassle.
Your options before:
Pieter van Dijk retweeted
🚀 DeepSeek V4.1 Flash 1017 now runs on just 2x DGX Sparks 🔥
TensorFold's 2-node engine + LibertAI's EXL3 2.9 bpw pack:
⚡ 73 tok/s for 1 user, 136 tok/s total at 4 users
👤 84 tok/s per user solo
📥 93K-token prompt read in 43.7s (2,135 tok/s)
🧠 Full 1M context, loads in 27s
2 Sparks now beat our old 4-Spark vLLM setup (48.5 tok/s) 😤
Full recipe, scripts + every number 👇
github.com/tonyd2wild/DeepSe…
Big thanks to bertholomus for the TP2 engine 🙏
🚀 DeepSeek V4.1 Flash 1017 now runs on just 2x DGX Sparks 🔥
TensorFold 2-node engine + LibertAI's EXL3 2.9 bpw pack: full 1M context, loads in 27s, read a 131K-token prompt at 2,059 tok/s
Speed numbers dropping soon 👀
Repo 👇 github.com/tonyd2wild/DeepSe…
Pieter van Dijk retweeted
We are pleased to share that the ADIPEC Strategic Conference agenda is now available.
Bringing together global leaders from across energy, finance, policy and technology, the conference will explore the priorities shaping the future of energy and the opportunities driving long-term progress.
Across 11 dedicated programmes, discussions will span energy security, investment, infrastructure, innovation and the evolving role of technology in supporting global growth.
ADIPEC is where leadership, insight and collaboration come together to shape the next chapter of energy.
Explore the Strategic Conference agenda: bit.ly/4zD0ZLA
Register as a delegate for ADIPEC, taking place from 2-5 November 2026 in Abu Dhabi: bit.ly/3StKHDV
#ADIPEC #ADNOC #StrategicConference #GlobalEnergy
Pieter van Dijk retweeted
New in agent-browser:
snapshot --delta
Returns only what changed since the last snapshot, instead of the whole page
Below: 73x fewer tokens filling a form
Pieter van Dijk retweeted
This article is worth reading... It will give you plenty of ideas on how to use Claude
Claude solved a 25 year old cryptographic mystery in under 30 minutes
Researchers had spent decades trying to understand how the NSA generated the original seeds for some of the most widely used ECDSA curves
A few interesting details:
- Claude identified a pseudorandom number generator from the 1990s
- Reconstructed its internal state using mathematical techniques
-Traced the seeds back to unix timestamps from 1997 and 1998
- Wrote code to reproduce and verify the original values
- All of this in less than 30 minutes
A few days ago, I faced a similar challenge
I needed to understand how an id returned by a service was generated. I had to decode it or find a join between database tables to identify the complete record
I searched everywhere and found absolutely nothing... but I knew the id had to be connected to the other records somehow
That's when I decided to run Opus 5.5 + Dynamic Workflow
This is exactly the kind of problem where Dynamic Workflow is useful. Instead of following a fixed sequence of steps, the agent can investigate, generate hypotheses, run experiments, evaluate the results, and decide what to explore next
After about 20 minutes, Claude came back with the answer
The id was generated by a python function that combined the record's name with other values to produce the identifier
No documentation, no obvious relationship between tables. Just investigation, hypotheses, and verification
I recommend using Claude for these kinds of challenges. Give it a difficult problem, the right tools, and room to investigate
You might be surprised by what it discovers
Frontier Red Team Link: anthropic.com/research/ecdsa…
Pieter van Dijk retweeted
Most developers aren't aware of this hack but you can actually run coding agents without any API keys and substantially reduce any risk associated with agents leaking keys.
This is generally a best practice and we implement this using a transparent proxy at @infisical.
The general idea is that you scaffold the agent's environment to proxy all traffic through a transparent HTTP proxy. This is non-intrusive and covers all CLI, local MCP, LLM provider traffic.
On a more practical level this means:
- OpenCode invokes the GitHub CLI to raise a PR without holding a GitHub PAT.
- OpenCode connects to any LLM Provider (e.g. Anthropic) without holding any API key.
- You get the point.
We've open sourced Agent Vault which holds agent credentials as part of Infisical and Agent Proxy, the standalone companion transparent proxy that fetches back credentials and brokers them for agents like OpenCode.
Linking the docs below; you're welcome 👇
Pieter van Dijk retweeted
Mohammed bin Rashid launches new cycle of Arab Hope Makers
The initiative will seek out people dedicated to doing good, serving humanity and creating hope in their communities.
Read more: emirates247.com/uae/mohammed…
#ArabHopeMakers #UAE
Pieter van Dijk retweeted
Check It Out >> The October Issue of the @CorixPartners Cyber Security #Leadership Newsletter
buff.ly/sabZHEU
A reference resource for the #CIO and the #CISO on #cybersecurity, looking beyond the #technology horizon into leadership, #management, culture & #governance
Pieter van Dijk retweeted
NVIDIA's Qwen3.8-Flash-Next on Spark 1m context
1 million tokens read from SSD (on the Spark) into KV-cache, TTFT just 8.5 seconds (5.5 directly from the cache) and then 65 tok/s!
(YaRN setting factor 4.0 over 262,144)
That's just CRAZY!
huggingface.co/nvidia/Qwen3.…
Pieter van Dijk retweeted
Amazon has sold over 500 million Alexa devices.
Most people use theirs for 3 things:
Timers. Weather. Music.
A smart home installer who has set up 1,000+ homes told me:
"Alexa is the cheapest home automation hub you'll ever buy. One sentence can turn off every light, lock the door, and set the alarm. People use it as a $50 kitchen timer."
He set up 10 routines in his own home.
They replaced $116/month in apps and services.
Here's every routine he uses 🧵











































