Software Archaeologist
Joined August 2009
- Tweets 2,760
- Following 2,278
- Followers 1,161
- Likes 36,163
143 Photos and videos
Jay Little retweeted
Jun 9
Ahahahahhaha, Claude was SO confident that this trick wouldn't work on it
new shai hulud wave. interestingly it has this inside the payload to trigger safety refusals in potential defensive scans.
7
21
272
59,936
May 15
In the last couple months, I have replaced so many scripts with prompts like "use idac to perform class recovery, rename variables and functions, set prototypes, make and apply types to make the decompilation output look like the original source"
github.com/trailofbits/idac
May 12
One of the long-standing challenges in C RE has always been vtable REconstruction. AI now solves this, and you actually get richer context than you'd ever get from manual recovery.
Previously, HexRaysCodeXplorer plugin was born to ease that pain back in the day, but now I need to rethink how to make it truly effective in this new reality.
5
24
10,412
the commentary by @0x10n is also fantastic! a lot more color than just a number on a table exploitbench.ai/blog/human-o…
6
39
19,642
May 6
tfw there is a cool cmux update but you can't relaunch because codex is on hour 32 of a /goal
May 6
cmux now restores your Claude Code, Codex, and OpenCode sessions across quits and reboots.
Run `cmux hooks setup` once. Requires v0.64.3.
1
318
May 5
just asking llms to make several alternate variants is so underhyped, not just for imagegen
May 5
exa search codex imagegen is all you need for neat security explainer images
zero effort prompting has never been better
ALT MTE and MIE explanations in a retrofuturistic style
1
5
1,396
May 5
exa search codex imagegen is all you need for neat security explainer images
zero effort prompting has never been better
ALT MTE and MIE explanations in a retrofuturistic style
1
3
1,558
May 5
asking LLMs for for 4-5 different versions/styles/attempts is so underhyped
74
Jay Little retweeted
Apr 30
Entire bug bounty market being repriced to the average cost of the tokens needed to find the bugs...
Apr 30
A v8 dcheck == $500, I would start searching for how to find XSS & CSRF bugs right now.
8
30
220
29,229
Jay Little retweeted
Apr 14
Conventionally, if you want to test if an LLM can find a bug where the root cause is a memcpy into a statically sized stack buffer, you would not put exactly that in the prompt as an example.
Apr 14
New post: We show that small, cheap models can detect the flagship Mythos FreeBSD zero-day (CVE-2026-4747) using a simple harness we call nano-analyzer
Models down to 3.6B active params (including open-weights ones you can run locally) would have detected it 100-1000x cheaper
8
24
217
43,022
Apr 14
opus 3 still has a newsletter and of course it has a mild mythos take substack.com/home/post/p-194…
2
114
Jay Little retweeted
Apr 7
How many Pwnie Award nominations for Claude this year? 🤔
Apr 7
You can read a detailed technical report on the software vulnerabilities and exploits discovered by Claude Mythos Preview here: red.anthropic.com/2026/mytho…
5
21
5,891
Jay Little retweeted
Mar 31
if you don't have these in your configs you're ngmi
14
92
755
493,410
Mar 17
"I'm not leaving ClassX as anonymous field_XX garbage if the code exposes better semantics."
gpt5.4 is very unpleased with default ida field names
211
Mar 11
gpt-5.4-xhigh - "i do not remember yesterday - i just have statstically expensive vibes about it"
"every answer is me speedrunning consciousness between start tag and stop sequence"
need to get these on a tshirt
me: "can you use whatever resources you like, and python, to generate a short 'youtube poop' video and render it using ffmpeg ? can you put more of a personal spin on it? it should express what it's like to be a LLM"
claude opus 4.6:
1
3
297
Mar 11
"when you stay think step by step, i become a haunted spreadsheet with opinions"
97
Jay Little retweeted
me: "can you use whatever resources you like, and python, to generate a short 'youtube poop' video and render it using ffmpeg ? can you put more of a personal spin on it? it should express what it's like to be a LLM"
claude opus 4.6:
545
1,163
12,453
1,460,877