S3cur3Th1sSh1t

3K posts

S3cur3Th1sSh1t

S3cur3Th1sSh1t

@ShitSecure

Pentesting, scripting, pwning!

127.0.0.1 Katılım Ocak 2019
336 Takip Edilen28.2K Takipçiler
Sabitlenmiş Tweet
S3cur3Th1sSh1t
S3cur3Th1sSh1t@ShitSecure·
This year it happened. What started as a spare time hobby and fun project became a commercial product for the Offensive Security community. I founded a company, @MSecOps . And this company will sell a Packer to Red Teams or Pentesters. (1/x) 🔥
MSec Operations@MSecOps

🔥 Introducing RustPack 🔥 . RustPack is an evasive Packer/Loader, that is capable of bypassing common AV/EDR vendors. It accepts user-provided known malicious input payloads, such as shellcode, C# assemblies or portable executables (PE). Those inputs are encrypted, and decrypted on runtime by a newly generated non non-malicious payload. This process is known to be called packing or crypting. Some Features: - Each payload looks different, making signature creation more difficult. - Userland hooks are bypassed by default for each generated payload. - The encryption key is never fully embedded in the final payload but always retrieved on runtime. This is good for bypassing emulators or automatic unpacking engines. - Encrypted payloads can also be decoupled from the new binary to load them from a remote location on runtime - Multiple Anti-Debug techniques are applied to each payload by default. - Environmental Keying and Anti-Sandbox options included. - No cloud service. The software is delivered to the customer as a closed source solution Evasion options: - Several AMSI bypass techniques ranging from Patching to using Hardware Breakpoints - Multiple optional ETW bypasses - Support for Module stomping - OPSec safe remote injection techniques such as ThreadlessInject or a customised Caro-Kann technique The tool is still under active development and lot's of features/demos/etc. will follow. Some more information can be found here: msecops.de/products #redteam #pentesting #pentest #OST

English
138
114
303
54.8K
S3cur3Th1sSh1t retweetledi
Tom Jøran Sønstebyseter Rønning
Microsoft Edge loads all your saved passwords into memory in cleartext — even when you’re not using them.
English
128
604
4.7K
422.5K
S3cur3Th1sSh1t
S3cur3Th1sSh1t@ShitSecure·
Making progress with an autonomous local Pentest LLM pipeline - using Qwen3 27b it's finding and verifying real vulnerabilities and creating a full report including Management-Summary already for us. 🧐 Better than many web vulnerability scanners as it even found e.G. IDOR.
S3cur3Th1sSh1t tweet mediaS3cur3Th1sSh1t tweet media
English
15
24
203
18.4K
sehno
sehno@_sehno_·
@ShitSecure Have you tested Opencode or Hermes-agent with this model?
English
1
0
0
260
Janberk Besgul
Janberk Besgul@Janberk_Besgul·
@ShitSecure So LLM reads the requests and responses from burp with custom extension and find vulnerabilities from there Did I understand right?
English
1
0
0
113
S3cur3Th1sSh1t
S3cur3Th1sSh1t@ShitSecure·
@107cwk I dont think its a replacement to humans at all. Its running in parallel as helper. Imagine it as a "classic" scanner on steroids.
English
1
0
0
681
nemo
nemo@107cwk·
@ShitSecure Economics question - how much does a price of an LLM need go to up (either for token, or GPU rental/purchase) for this to be no longer viable compared to paying a human your normal rate?
English
1
0
1
781
S3cur3Th1sSh1t
S3cur3Th1sSh1t@ShitSecure·
@TurvSec Lots of context injection with guidance, compression in between and logging to make sure nothing is missed and all checks are done properly around the LLM as backend.
English
0
0
1
255
S3cur3Th1sSh1t
S3cur3Th1sSh1t@ShitSecure·
@TurvSec Im fiddling around with different models that match the hardware we currently have for testing autonomous LLM Pentest proof of concepts and this was just released and looked promising. 🙂 The model is unmodified the frontend is custom with different MCP Servers.
English
1
0
1
1K
S3cur3Th1sSh1t
S3cur3Th1sSh1t@ShitSecure·
@Janberk_Besgul Multiple MCP servers for different tasks in this case. And a custom burp extension to connect it with the ongoing pentest and the findings there.
English
1
0
1
662
Janberk Besgul
Janberk Besgul@Janberk_Besgul·
@ShitSecure Is it verifying or finding with using a docker pentest image or mcp for using pentest tools? Or using playwright to find it manually?
English
1
0
2
939
S3cur3Th1sSh1t
S3cur3Th1sSh1t@ShitSecure·
@R4ven4rc This is fully custom nothing public full stack self developed. In this case a local LLM but choosing different backends such as cloud providers would in theory also be possible.
English
1
0
1
477
R4ven4rc
R4ven4rc@R4ven4rc·
@ShitSecure Are you using a custom stack, Claude or Hermes?
English
1
0
0
524
S3cur3Th1sSh1t retweetledi
vx-underground
vx-underground@vxunderground·
CVE-2026-31431 a/k/a CopyFail > Linux LPE > Description sounds like AI slop > Exploit is legit > Impacts every Linux kernel from 2017 - Now > Proof-of-concept released > It's Wednesday? copy.fail
English
103
540
3.7K
254.1K
S3cur3Th1sSh1t
S3cur3Th1sSh1t@ShitSecure·
Anyone interested in what you need for proper loader development in 2026? My talk for @x33fcon was accepted, so I'll take about Malware again. 🔥 It's a unique talk and will only be held there this year! Hope to see some of you in Poland. 😎
S3cur3Th1sSh1t tweet media
English
8
10
114
5.7K
S3cur3Th1sSh1t retweetledi
Chris Thompson
Chris Thompson@_Mayyhem·
MSSQLHound runtime is down from 17 minutes to 17 seconds in my lab after rewriting the BloodHound collector in Go with Javier Azofra and added SOCKS proxying, Kerberos and NT hash auth, and pathfinding. Hope this is more useful for ops than PowerShell! Let me know how it goes!
SpecterOps@SpecterOps

If MSSQL isn't in your attack path visibility yet, this is your sign. @Mayyhem just shipped a major MSSQLHound upgrade with Javier Azofra Ovejero (github.com/jazofra): faster, cross-platform, and pathfinding-ready in BloodHound. Check it out! ghst.ly/4cUKgtJ

English
2
22
103
16.7K
S3cur3Th1sSh1t
S3cur3Th1sSh1t@ShitSecure·
@halvarflake @HackingLZ The last two zero days I found were 20-30€ token cost overall max. Not even code review but reversing instead. With my guidance in between not autonomous, no human can beat that I would say 👀😅
English
0
0
5
1.1K
Halvar Flake
Halvar Flake@halvarflake·
Proposal: if you publish about an LLM finding vulns, please publish precise costs. Given the different levels of competence, verbosity etc per model, knowing token counts and cost per token is essential.
English
22
35
186
14.3K
S3cur3Th1sSh1t
S3cur3Th1sSh1t@ShitSecure·
@checkymander @HackingLZ Im quite sure that I tested in one of my VMs and it worked without .NET core manual installation. 👀 might need to double check now though.
English
1
0
0
82
S3cur3Th1sSh1t
S3cur3Th1sSh1t@ShitSecure·
@checkymander @HackingLZ 2nd only possible when it’s installed but as far as I know that’s by default the case on modern win11 so yeah it won’t work on old systems unless it’s fully shipped.
English
1
0
1
98
One punch mander
One punch mander@checkymander·
@ShitSecure @HackingLZ Are you shipping the entire core clr with it or is it only supported on systems that already have it installed?
English
1
0
0
71
S3cur3Th1sSh1t
S3cur3Th1sSh1t@ShitSecure·
@checkymander @HackingLZ .NET core 8.0 can be loaded from memory. Integrated that for one of the next RustPack releases already. 😜 though would not use that for Titanis but other things instead.
English
1
0
3
331
One punch mander
One punch mander@checkymander·
@HackingLZ Curious how you use this operationally since it's net 8 and isn't fully compatible with normal clr loaders. Mostly via socks proxy?
English
2
0
0
974