c4ch3c4d3

🚨 RTVCron Ep. 5 is coming 📅 Sat, May 9 at 3PM ET 🎙️ @c4ch3c4d3 "Local Large Language Models" Learn how to run LLMs on your own hardware, what works, what doesn’t, and how to make it happen on any budget 📺 Click “Notify Me” so you don’t miss it youtube.com/live/L8bYnzq4J…

@JaidCodes @UnslothAI @Alibaba_Qwen Use the nightly build, it works fine.

@UnslothAI @Alibaba_Qwen looking forward to test it in vLLM v0.170 in 2034

You can now fine-tune Qwen3.5 with our free notebook! 🔥 You just need 5GB VRAM to train Qwen3.5-2B LoRA locally! Unsloth trains Qwen3.5 1.5x faster with 50% less VRAM. GitHub: github.com/unslothai/unsl… Guide: unsloth.ai/docs/models/qw… Qwen3.5-4B Colab: colab.research.google.com/github/unsloth…

@Steph3nSims I've had some success with using locally hosted LLMs & agents to automate cloud / network enumeration tasks. Its squirrely, but there's lots of opportunity to explore data troves that I'd normally have limited time to dive into.

For those of you working as a pen tester, red teamer, or other offensive cybersecurity role, are you using autonomous agents? I'd love to hear your comments as to what you're using, for what, any limitations, and if not, why? Thanks!

@0xTib3rius With only 12gb VRAM, you can chat, but any agentic flow will be slow. Depending on how much memory you want available for active chat: GPT-OSS 120b gets you smarter answers, but little overhead GLM-4.7-Flash gets you long convos Qwen-3-Next is a possibly a good middle ground

What are the "best" general purpose local LLM models people are using these days? Specially stuff that could run on a PC with 64 GB RAM and an RTX 4070?

Could Your Employees Be the Weakest Link in Your Security? opensecurity.com/could-your-emp…

@evilsocket @noperator It has to do with how llama.cpp vs other inference engines like vLLM work under the hood. 5 inch deep answer - vLLM with the --tensor-parallelism allows you to utilize your GPUs compute simultaneously, vs. ollama & llama.cpp which don't a similar feature.

@noperator Oh that’s interesting, I wonder why

🚨 DC33 calls are still OPEN: Sponsors, Workshops, Volunteers! PLUS…Meet this year's TRON-themed mascot 👾 Help us name it! 💡 Drop your best name below ⬇️ and win a prize if chosen! 💥 Prize for best name 💥 Prize for 1 lucky like + retweet Winners must be attending #DC33!

🚨Special Announcement 🚨 Red Team Village has elected a new Board of Directors! 🎉 Please join us in welcoming: @MikeHacksThings, @ds1nk, @c4ch3c4d3 ,@forktheplanet, @briansak,@infosecoprah, @cwolff411, @FilipiPires, @th3d00rman A huge thank you to our outgoing board members who helped make RTV what it is today—we appreciate all your hard work! 🙌 RTV is excited for all that’s ahead!

@TJ_Null If you're going to run with Proxmox, consider giving Ludus a shot, you'll get some sick lab automation, and still can run your typical workflow with a little understanding.

Debating to wipe my test server that has VMware ESXi on it and I plan to install Proxmox. Any objections?

Joining Optiv back in 2020, Joey Belans was a huge mentor for me and helped me grow a TON. I will always be grateful for his kindness. Joey was let go today from Optiv just before starting FMLA, and his family could really use support. gofund.me/073fa52c

@MikeHacksThings @RedTeamVillage_ Took the wife there last year, so good!

Core team dinner with @RedTeamVillage_. Been wanting to eat here forever, hope it doesn’t disappoint!

@0sm0s1z At minimum, needs more cowbell.

Been tinkering with my “ultimate desk” since the move… I think it’s done? If done is even a possibility.

One of our amazing platinum sponsors is hiring! Check out these open roles at @amazon! 🤓 amazon.jobs/content/en/tea… #InfoSec

@0sm0s1z My strategy has always been to just reformat after getting a new job 😉

Been craving a new computer recently and I’ve come to the realization that what I actually want isn’t an upgrade it’s just a boring, clean computer… Apparently I can’t clean up after myself 😅

I normally restructure my digital layout every time I replace my computer, but my M1’s lasted so long that I’m becoming an utter mess. I can’t find anything. Disaster.

@0xTib3rius Ironically, Microsoft seems to disagree with this sentiment based on the newest version of Windows.

OK so I see I didn't convey my idea very well at all. I am *NOT* saying we should record our screen the entire test and then edit it down!!! This would be the process: 1. Find an exploit where you think a video conveys more information to the customer than a series of screenshots. 2. Start recording. 3. Perform the exploit. 4. Stop recording. 5. Attach the video to the report. It's literally the same process you would do when you take screenshots, only you are recording the entire exploit. You could even then just extract screenshots from the video for the PDF report.

Is it time to normalize more interactive pentest reports? A PDF is nice, but there are issues that are better presented via a short video (not even necessarily with audio) than a series of screenshots with text. Are there are reporting platforms out there that support this?

@vxunderground Once we receive the shirt, what password will be needed to put it on?

In less than 24 hours we will officially be 5 years old. We are giving away $1,000 worth of our "5 horsemen of the apocalypse" shirt. If you'd like a free shirt leave a comment below.

@badsectorlabs Ludus is awesome! Used it to stand up GOAD & a personal elastic stack I use for testing already. Phenomenal project! Much Thanks!

Stop testing in prod (even someone else's)! Are you tired of installing Active Directory on your test VMs for the 100th time? Ever YOLO a binary off GitHub into prod because your testing setup is tedious? I've built a solution: ludus.cloud (1/5)

@bettersafetynet I'm an immense fan of perplexity.ai, cuts out much of the chaff a la Bing search, but is actually good

Given how bad google's search is... how soon is it before someone finds a way to have workable search engine content again? I'm assuming you'd have a massive market capture overnight if you could crack this. It shows they're an ad delivery platform. :-/

@0sm0s1z There are plenty of cheesy old systems that have an applicable module out there. My biggest shout outs are JMX and IPMI hash disclosure.

So… real talk. How often do we actually fire up Metasploit these days?

@TimMedin Pizza chef! I've been working hard at getting the perfect crust out of my backyard oven.

If you didn’t do Cybersecurity, what would you do for a career? Why? Follow up, what if you excluded tech?