Chris Nickerson retweeted
cvENGINE: the most advanced external attack surface assessment you can get for free. Point it at a domain: bbot, nmap, and Nuclei map your attack surface, then an AI proves which CVEs can actually hit you with read-only checks.
github.com/hevnsnt/CVEngine
Chris Nickerson retweeted
NVIDIA ACABA DE ELIMINAR EL MAYOR PUNTO DE FRICCIÓN EN VOICE AI
han liberado PersonaPlex-7B, un modelo de conversación dúplex completo que puede escuchar y hablar al mismo tiempo
→ no espera a que termines de hablar
→ usa una arquitectura de doble flujo para procesar y responder en tiempo real
→ código disponible en GitHub
y es 100% open source y gratis
os dejo el repo abajo
Chris Nickerson retweeted
I highly recommend cloudflare's security audit skill: github.com/cloudflare/securi…
Run it on all your software. Run it with the free models. Use it even if you used a model to audit something - it will find much more things. Works in all agent harnesses and is model independent.
Chris Nickerson retweeted
Introducing tsrs: a Rust port of the TypeScript compiler 🦀
We are not forking TypeScript!
◆ In close collaborating with the TypeScript team
◆ Up to 8x faster than TS 7.0 & ~70% less memory
◆ Runs in WASM, so it works in your browser & on Workers
◆ Soon part of the Oxc toolchain
npx tsrs
blog.cloudflare.com/typescri…
Chris Nickerson retweeted
🔓 Open-source release
Get DeepSeek-V4.1-Flash-1017:
🤗 Hugging Face: huggingface.co/deepseek-ai/D…
🔗 ModelScope: modelscope.cn/models/deepsee…
4/4
🏴☠️ Ahoy InfoSec practitioners! The CFP for #BSidesSF 2027 is OPEN!
Got research, a talk, or hands-on workshop to share? Bonus points if you tie it to our theme: Pirates! 🦜
📅 Deadline: Nov 9, 2026
📍San Francisco: April 3 & 4, 2027
Submit here👇
pretalx.com/bsidessf-2027/
Chris Nickerson retweeted
THIS IS ACTUAL F*CKING GOLD
Anthropic just shared one of the best tricks for Claude Code.
The idea: run Sonnet, Haiku and Opus 5.5 together, each one doing only the job it's best at.
Opus-level judgment on every big call, without paying Opus rates for every routine turn.
You just launch Claude Code like this:
claude --model sonnet --advisor opus
here's the split:
→ Sonnet 5.5 runs the session and writes the code
→ Haiku 5.5 explores files and docs in parallel
→ Opus 5.5 stays on call as advisor for the big calls
→ you stop spending Opus on work that doesn't need it
Opus only steps in at three moments:
→ before a plan locks
→ when the same error shows up twice
→ before calling it done
Sonnet builds. Haiku explores. Opus reviews.
JEV engineering is the same move one layer down: the forks that need no thinker (which file, which tool, retry or stop) go to JEV in milliseconds, and the big models only see the ones that split
the full tree
> Sonnet 5.5 runs the main session and edits code
> search maps the repo on Haiku 5.5
> grep finds call sites on Haiku 5.5
> docs pulls the API reference on Haiku 5.5
> Opus 5.5 on call as advisor, reads the whole transcript when consulted
> JEV routes micro-forks in milliseconds
paste the tree and this prompt into Claude Code ↓
"Rebuild my Claude Code setup around this tree:
1. Run the main session on Sonnet 5.5 with Opus 5.5 as advisor.
2. Create three read-only Haiku 5.5 subagents: search (maps the repo), grep (finds call sites), docs (pulls the API reference).
3. Add one rule to CLAUDE.md: consult the advisor before locking a plan, after the same error twice, and before calling a task done.
4. Route fast deterministic forks (tool choice, path selection, retries) through JEV.
Show me every change as a diff first. No edits until I say go."
↳ code.claude.com/docs/en/advi…
save it, you'll use this every single day.
Chris Nickerson retweeted
"I don't have a GPU" is officially dead 🤯
You can now run 70B model on a single 4GB GPU and it even scales up to the colossal Llama 3.1 405B on just 8GB of VRAM.
AirLLM uses "Layer-wise Inference." Instead of loading the whole model, it loads, computes, and flushes one layer at a time
→ No quantization needed by default
→ Supports Llama, Qwen, and Mistral
→ Works on Linux, Windows, and macOS
100% Open Source.
Readers added context they thought people might want to know
AirLLM enables the claimed low VRAM usage via per-layer disk streaming but requires reloading the full model from storage for each token, yielding speeds like ~20s/token for 8B models and minutes per token for 70B+ per tests and project reports.
kunalvohra.com/blog/airllm-on…
github.com/lyogavin/airll…
Chris Nickerson retweeted
Platforms charge $700/mo for options data.
We built our own GEX & VEX exposure Heat Maps...and made them cost $70/mo.
Built by a team from Nasdaq, Amazon, and Microsoft. Licensed exchange data. 6,000+ tickers, real-time.
1/10th the cost. 10x the capability. 7 days free.
China f*cked the world again
DeepSeek just dropped a killer library that saves up to 98% on tokens ur GPT Astra and Opus 5.5
it's DeepGEMM, the engine DeepSeek trains and runs its own models on
faster GPU math -> more tokens per GPU -> cheaper tokens
> plug the engine DeepSeek uses into your own models
> squeeze more tokens per second out of the GPUs you already pay for
> match or beat expert-tuned GPU libraries on speed
u will pay $1.6 instead of $50
its sitting in one public repo, link below
Readers added context they thought people might want to know
DeepGEMM is a low-level CUDA kernel library of optimized GEMM ops for custom inference engines on specific NVIDIA GPUs, mainly for DeepSeek's own MoE models. It cannot be plugged into proprietary models like GPT/Astra/Opus for claimed 98% token savings or $1.6 pricing.
github.com/deepseek-ai/De…
agentpedia.codes/blog/deepgemm-…
Chris Nickerson retweeted
Dissecting Impacket. public reference of protocol-level and implementation-level indicators of compromise(IoCs) for detecting Impacket-driven activity #DFIR
github.com/ThatTotallyRealMy…
Chris Nickerson retweeted
The best part about the Exposure Forecast tool is that you can go back and backtest 👇
🚨 Here is the exposure forecast for $SPY with price currently in volatile market conditions.
There is currently a 94% chance that volatile conditions remain in the next 30 minutes with a 0.14% move.
Chris Nickerson retweeted
harnessrouter/harnessrouter: HarnessRouter Community Edition: the self-hosted, Apache-2.0 edition of the unified interface for agent harnesses. Run Codex, Claude Code, Hermes, PI, DSH, and more through one API, with sessions, s... github.com/HarnessRouter/har…
Seems like the cartel's need to go "legit" and rent out their gpu's for everyone else to solve our hard problems. "Mexican cartel's crypto farm seized in mountain raid — 300 GPUs" share.google/Emk8e06lDReuZiP…
Chris Nickerson retweeted
Dear @AWS Bedrock,
I need to start seeing other endpoints.
You said you were model agnostic. I didn’t realize that meant indifference to which year they came from. Qwen, Kimi, DeepSeek, and MiniMax have moved on generationally. Meanwhile, we’re browsing your catalog like relatives clearing out an attic: “Surely someone can still use this?”
And then there’s Claude. You keep announcing exciting new versions while our account remains on Opus 4.6. Our upgrade path appears to be a waiting room with a press office.
I’ve tried talking about this. Support knows. Sales knows. Executives know. Our complaint achieved global availability before our model access did.
I’ve also seen you making time for OpenAI and Grok. So you can move quickly. That was useful information to discover this far into the relationship.
Our investors funded an AI company. Somehow, we’ve become a support-ticket management platform with an inference dependency. There’s probably a Gartner category for that, but I’d rather have a product.
I needed current models, competitive economics, and the ability to ship. You offered me another meeting with someone who would take this back to the team. At this point, I assume “the team” is a dead-letter queue.
Please don’t send a solutions architect. We’ve done couples therapy.
I’ve put your things in an S3 bucket. The bucket exists, your permissions are correct, and access is generally available.
If you can’t retrieve anything, please contact Sales.
Formerly yours,
A customer with a shipping deadline
Watching meta's Muse beat captcha and human verification live in the browser is truly making me smile today.
Chris Nickerson retweeted
In 1986 a guy got kicked out of every casino in Vegas for counting cards. So he flew to Hong Kong with $180,000 and started betting on horses instead. He walked away with almost $900 million.
It's Bill Benter. He figured horse racing was just another counting problem. Same math, more moving parts.
He and a partner showed up with $180k and a computer. Benter spent years teaching that computer to guess one thing, the real chance each horse had to win. If his number was better than the odds the bookies gave, he bet. If not, he skipped it.
That's the whole trick. Expected value.
EV = p · b − (1 − p)
Only bet when your win chance p, at odds b, is worth more than your chance of losing.
This recording was never meant to be some hidden gem. Nobody expected Professor Tsitsiklis to hand the whole foundation away in 45 minutes, but that's exactly what happens on the board. Students in that room pay over $80,000 a year to sit through it. It's free right here. It's free right here.
Every quant, every professional bettor, every hedge fund analyst started with this exact hour. Benter just watched it and actually did the homework.
Almost nobody knows this lecture even exists. Watch it before it gets taken down.
The answer is in this video.
Chris Nickerson retweeted
Made a Cybersecurity Resume Reviewer AI skill that I've been using for some mentees, made it public for anyone who is interested in using it as well. Hope others find it useful (also open to feedback or pull requests). github.com/mubix/cyber-resum…
Chris Nickerson retweeted
My top cyber influences and mentors in rough order:
NON-REAL
* David Lightman from WarGames - I actually wrote a fake NORAD I hacked on my Tandy Color computer and scared the crap out of my friends when I was like 8.
* Martin Bishop from Sneakers - Leadership, adaptability, going beyond the keyboard.
* Harry Caul from The Conversation - Paranoia, craftsmanship, navigating ethical dillemas.
REAL-WORLD
* Fravia & +ORC - I put them together because I discovered them on Friavia's site. I learned assembly / disassembly, patching, how to use the internet for intelligence, a whole mindset. HUGE impact.
* Optyx - He was the masked hacker on Tech TV and wrote the first MD5sum collision tool. At one point he had over 100 OSX 0days. He would send me exploits that were either just diffs or somewhat broken to get me to learn. He taught me how to really mess with IDS's by byte encoding / manipulation in network traffic. My kid drew the cartoons for his Kernel Intrusion System (KIS) which I think was the first linux rootkit publically released. He directly taught me a ton.
* HDM - Not enough words to say how much I learned from him. Actual development processes, how to write tighter smaller code, how to be humble while also being a bad ass, how to encourage people who are trying to learn and contribute, how to take no shit, all kinds of things about systems. How to give talks. One of the best out there.
* Halvar - How to find bugs in software. How to think in graphs. Why math is important. The whole idea of comparing patches to find 1days. Core concepts.
* Ramune - No one has heard of him but he could answer any question on any subject of computing no matter how obscure. Any OS, programming language. Just a genius.
* Mark Russinovich - I've never met or talked to him, but his Windows Internal books and Sysinternals tools form a core of my whole career.
* Pedram Amini & Ero Carerra - I put them together because I met them together. Malware RE, file formats, patching the Z flag to mess with anti-analysis branches. HUGE influence. Pedram gave me a customized Ollydbg which I probably used for 12 years daily.
* Jon Erickson - I had to buy a second copy of his book I used to first one so much. This is where I really learned memory corruption concepts and return to libc (later ROP). Also he was a super nice guy to me.
* Ed Skoudis- I read his Malware book probably 10 times. Like HD, he really modeled kindness and openess in a community that can often be elitist and mean.
* Mudge - Buffer overflows, DARPA. I remember him giving an almost Terrence McKenna like talk I think at defcon which showed a very unique way of thinking about security. Impactfull guy.
* Jamie Butler & Greg Hoglund - Met together. They taught me how to live in and abuse the kernel. I tried to replicate everything in their rootkits book. HBGary's Responder opened up my mind to the idea of automating and productizing the stuff I was doing manually.
* Parmaster - Oldschool blackhat. Taught me about infiltrating hacking groups, how to be a gangsta (he would roll into vegas in a velour track suit and sunglasses). He was generally a good mentor. He was also in about every secret society known to man so I learned about the Masons, Rosicrucians and others from him.
* Joanna Rutkowska - I didn't know her well, just met a few times, but her command of the kernel and memory was revolutionary to me. I once saw her give something like a 300 slide talk at Blackhat and it was the most deeply technical and fastest spoken talk I've ever seen.
* Dino Dai Zovi - Super cool, nice, and humble dood. Wifi hacking, continuation of execution, reliability in exploits, he's an artist. He came up with a hilarious volcano eruption meme at a conference that a bunch of us copied. I regret that we never got to work together but I learned a lot from his work.
* Sangfroid - I learned less technical stuff from him, but rather strategic thinking, business, culture, and how to keep what is most important in focus.
* Ac1db1tch3z - How to be scary. The idea of supply chain compromise. Mad respect.
* Hobbit - How to go barefoot and be a brilliant curmudgeon and why I don't hate javascript enough. Seriously though I'm not sure a day has gone by since it came out that I havent used netcat. He would walk around barefoot, with a screw driver shoved through his belt, wearing cuttoff jean shorts and a pony tail and I thought he was the coolest guy ever.
* SirDystic - He could be mean at times, but I got to help a little and learn from him on SMBRelay. The idea of protocol attacks was completely foreign to me and it took me embarrassingly long to understand wtf he was doing. I also watched in IRC/#cdc as he built BackOrifice which blew my mind and got me very into malware / C2. I don't think he liked me much but I learned a lot from him.
* Skape & Sp00nm - I was on a private SILC chat with those guys for years and I got them to come out and give my team a training class. Meterpreter was one of the coolest things I ever saw and I used it to hack ~100k computers. Skape was super down to earth. Sp00nm taught me some ASM and the concept of assembly golf. Really nice, brilliant guys.
* Chris Eagle - I had never even heard of the concept of emulation when x86emu came out. Blew my mind and became a huge part of my workflow. He was also super kind and open.
* Solar Designer - What can I say, dood is an alien. I'm convinced he might be the smartest / best hacker in the world, but he's so quiet about it. I used John the Ripper constantly for 20 years and there are mysteries about it I still don't know / understand. Never met him sadly.
* Richard Johnson - Brilliant guy. One of those people who would quietly msg me on IRC and point me in the right direction. I learned all about fault injection from him, lcamtuf's AFL, and Dave Aitel's SPIKE. I'm very thankful for knowing Rich.
* Rain Forest Puppy - I was a network hacker primarily but RFP's work opened me up to web apps. I didn't know him well, met once or twice, hung out on some chat servers. Revolutionary guy.
* Johnny Cache - I once saw him singlehandedly win defcon CTF against a bunch of whole teams of people. I focused a lot on the mechanics of actual hacking into systems and he was one of the best I ever saw. Also super nice and funny guy.
* Illfak Guilfanov - I never knew him, but IDA is the goat and I've used it most days since it came out.
* Phrack - Of course is on the list. If you don't read it, are you even really a hacker?
* The Core Impact Team, Especially Max - The concept of Pivoting and Syscall proxying had a massive impact on me. I used it in my own tools later on. I heard "Level 1 Agent Deployed" 2934723984723 times lol.
I know there are more so sorry if I missed you. These are the ones that came to mind as I was writing this. Thank all of you for the teaching and impact you had on my life and career.
Babe, we have VirusTotal at home.
Introducing POETRIAGE, LLM assisted malware triage.
It creates a sandbox environment for an agent to analyze and triage files, and generates a report based on the results.
Check it out here:
github.com/Vsimpro/poetriage

























