Haris Andrianakis

Founder, ex-Google, Information Security

[lcamtuf @lcamtuf]

Did you know that before LLMs, people sometimes actively sought "knowledge" and turned to so-called "books"?

If you want a sense of how our ancestors lived, "The Secret Life of Circuits" is shipping in just a few weeks: https://t.co/M5ZnXuT8js

ToC & sample chapter included.

[Haris Andrianakis @tl0gic]

@lcamtuf Two weeks of tokens?? lool

[lcamtuf @lcamtuf]

Looks like the latest edition of Paged Out! is out. This one features an article by yours truly, along with dozens of others:

https://t.co/ttcRAZ34Ti

[Haris Andrianakis @tl0gic]

@XorNinja What if all the coding agents running on engineers laptops went rogue?

[ClaudeDevs @ClaudeDevs]

Try Opus 5 today. Use “claude-opus-5” on the Claude API, or run “/model claude-opus-5” in Claude Code.

https://t.co/Z4WkWCSoGY

[Haris Andrianakis @tl0gic]

@lemonsqueezy Why is store activation taking so long? How can we get an update? Your support website says 2-3 business days but it's been over a week.

[Haris Andrianakis @tl0gic]

@dvassallo @jasonfried Login can be highly satisfying

[Calif @calif_io]

Reverse engineering Apple’s silent security fixes, by @blacktop__

We grabbed the latest iOS update, and diffed it with ipsw. The diff reveals at least two security-relevant changes that were shipped quietly.

https://t.co/mUt11Qy7ly

[Haris Andrianakis @tl0gic]

@dvassallo @GergelyOrosz +1, I'm wondering that too. I had already opted-out before this email so they did retain my setting.

[Haris Andrianakis @tl0gic]

@tdinh_me @ComprestoApp They must be using the The Pied Piper algorithm :)

[Haris Andrianakis @tl0gic]

@lunchbag Couldn't have said it any better -> "we want to truly be our own bosses and work on our own schedules." 🙌

[Haris Andrianakis @tl0gic]

@lcamtuf Who needs a password manager when you can ask AI to remember all your passwords for you?

[Haris Andrianakis @tl0gic]

@bcherny I see it in Claude Code 2.1.32, thanks!! Can't wait to try it.

[Boris Cherny @bcherny]

I've been using Opus 4.6 for a bit -- it is our best model yet. It is more agentic, more intelligent, runs for longer, and is more careful and exhaustive.

For Claude Code users, you can also now more precisely tune how much the model thinks. Run /model and arrow left/right to tune effort (less = faster, more = longer thinking & better results).

Happy coding!

[Haris Andrianakis @tl0gic]

@matt_elevenlabs 11B

[Haris Andrianakis @tl0gic]

@carste1n In my experience LLMs are better in some languages than others. But generally speaking you need to do some hand-holding and also to understand the codebase a bit.

[Haris Andrianakis @tl0gic]

AI: This seems broken, I can't figure out why. Let me reimplement the whole thing in a different way.
me: Can you add some debug logs to this external library?
AI: Ah I found the bug, it was a one line fix.

[lcamtuf @lcamtuf]

Fun computer fact: in binary data streams, zeros don't convey information and are included solely for padding purposes. They can be removed to make files smaller, then inserted back if a particular program requires padding for compatibility.

[Haris Andrianakis @tl0gic]

@carste1n @lcamtuf 1

[Haris Andrianakis @tl0gic]

@dvassallo Im happy with the Shark but I cant say the plastics arent cheap on those as well. Also never had a Dyson to compare ;)