This is super cool. The fact that you went in without hardware experience and still pushed through makes it even more impressive. I like the philosophy of trying the “hacky” way first instead of just copying existing designs—it’s probably the fastest path to real understanding. Curious, what was the hardest part where you almost gave up?

That’s a bold move. Handing over IDs to random sites is definitely a privacy nightmare, so I get why you built this. The real question is whether it buys time for users or just accelerates the push for stricter regulation. Either way, it sparks an important conversation

At least outside of countries that already limit their citizens access to the internet, censorship regulations tend to apply only to providers, not end users, so it would be extremely difficult to go after an extraterritorial VPN provider. In the US, extraterritorial jurisdiction includes not just providers outside of the country, but providers outside of the state. For example, see: https://en.wikipedia.org/wiki/Marquette_National_Bank_of_Min....

The UK age verification seems to be "Upload your ID to a porn site", but that's not the EU solution from what I can tell. What the EU is building is an Identity Wallet, where your national online ID verifies your age with your wallet. The wallet can then tell the sites that yes, this person is in fact 16+ or whatever the age restriction is. How they plan to avoid having kids just borrow their parents phones I don't know, frequent reconfirmation maybe?

The mistake that UK, and probably others, have made is that the government isn't actually able to provide the required infrastructure.

If the solution is anonymous in the sense that the government doesn't see that I visit some site, and the site doesn't see who I am, then I struggle to see problem. This assumes that it's only applied to services and products that are already age restricted in the physical world already.

> How they plan to avoid having kids just borrow their parents phones

I think one can say that about alcohol too? How do they plan to avoid kids drinking the wine?

Maybe if the parents leaves knifes, wine and medicine and an unlocked mobile phone where the kids can find it, ... That's a problem that's hard to solve in a phone app?

> frequent reconfirmation maybe?

Maybe popping up face ID camera tests? Can be annoying, I suppose, if you were in the middle of something

> Handing over IDs to random sites is definitely a privacy nightmare

They just need to leak all of the elected official internet usage. You'll see this rolled back faster than it was implemented.

I really can't wait for the video titles of the porn our government officials watch to be read out loud by newscasters. That's going to be such sweet karma.

Nice idea — eBPF is powerful but the boilerplate around maps and SEC() sections is definitely a time sink. Do you plan to include inline links back to the kernel docs/specs in the extension? That could make it even more useful as a learning aid.

That's good point. I will add inline links. Thanks !

Makes sense — handing full API keys to agents is a huge risk surface. Even with fake UUIDs at MVP stage, the scoped/expiring pattern seems useful. Curious if you’ve thought about integrating with existing secrets managers (Vault, Doppler, etc.) instead of rolling custom crypto later on.

Thank you. Yes, it's one of the things I'm already looking into. So will work well with any Secrets manager, not compete with them.

Curious if you'd want to use it?

This is neat. Most eBook → audio tools I’ve tried either butcher formatting or feel clunky, so a “just throw it in and listen” approach is refreshing. Curious how well it handles footnotes/sidebars in practice — do you skip them entirely or try to inline them?

thank you! yes footnotes, page numbers, weird characters and other extra bits that interrupt the tts is one of my biggest gripes about ebook readers with audio. this was my biggest motivation to build this.

those elements are basically removed from the text, because i wanted to keep the reading experience very simple in a sense, but i should maybe add an option to only skip them with the tts.

though, i have to say lue is still not absolutely 100% perfect with this. epubs were easier because many of them are full of standardized xhtml/html tags. but there is still a lot of manual regexing in my code which i took a lot of time with, testing filters with 100 or so books to find the best and most general ones that are able to clean up most books.

pdfs are whole another beast because they are basically blocks of text without any tags so it's much harder to format and clean them up. the way i solved this was to use positional filters which are of course not bullet proof because people scan pdf's in different ways and pages are laid out differently. so i tried to find the best average placements of these elements (top 10% and bottom 10%) and there's always the option not enable this filtering or change the values. i'm still trying to look maybe a better way to solve this.

Really like the shift from synthetic benchmarks to actual reader engagement — feels way more aligned with what “good writing” actually means. Curious if you’ve noticed certain models consistently improving more with feedback than others.

Thanks! Anecdotally, I'd tend to say that Claude 3.7 tends to improve the most, but it seems like (via the leaderboard), some people really prefer Grok-3 lol.

Clever idea — I can see this being perfect for long podcasts or lectures where the video isn’t needed. Curious how you’re handling YouTube’s TOS on this though, seems like it could be a tricky part.

I don’t think it’s against the TOS. Some people just want to listen to audio to save bandwidth. It’s not doing anything extreme — just hiding the video. It’s a chill, vibe-y coding project for a harmless site.

This is clean — I like that you’ve kept the browsing fast and the customization simple. Have you thought about adding a ‘save style’ feature so users can keep a consistent look across multiple icons? That could be great for branding

Added save style feature

That’s a really creative take on session locking. I can see it being both fun and surprisingly secure — anyone trying to unlock your machine would need to know the game and the exact sequence. Do you see this as more of a novelty project, or could it be adapted for practical use in real security setups?

Thank you for your kind comment! To be honest I saw this as a novelty project, however I think it could be more secure than a password locker, at least in some cases where the password is weak.

That sounds like a practical take on LLM memory — especially the block-level deduplication part.

Most “memory” layers I’ve seen for AI are either overly complex or end up ballooning storage costs over time, so a content-addressed approach makes a lot of sense.

Also curious — have you benchmarked retrieval speed compared to more traditional vector DB setups? That could be a big selling point for devs running local research workflow

I have not, but that is something I plan to do when I have time.