Feed aggregator

Show HN: Recursi – self-improving LLM-connected coding environment

Hacker News - 6 hours 5 min ago

A coding environment designed to be "recursively self improving".... but a whole lot more. Uses web based chatbots to save tons of money while being within terms of service and still being efficient. Runs right off the web (with no signup) and saves to your browser and/or straight to disk. Open source. Lots of cool sample/template apps.

Comments URL: https://news.ycombinator.com/item?id=48401022

Points: 1

# Comments: 0

Categories: Hacker News

Notes about a random free project I did 30 days ago (yt video transcriptions)

Hacker News - 6 hours 6 min ago

So about one month ago I did setup a fairly simple youtube transcription website, I have been vibe coding it for around 2 days and shipped the initial version;

The whole thing about it is because I want to crack the SEO game once for all and Im changing my strategy by offering useful tools to people.

All my previous attempts were focused on making the next useless blog where i generated articles with AI; that Google would ban after one week.

After 30 days my website is still indexed and live on google, and for now I consider it a win.

But besides the other projects that I built in past years, this is actually being used by people; for this Im really glad some people use the website by a random guy on the internet, its not exorbitant usage at the moment I get around 5 to 10 youtube transcriptions per day.

So the cool thing that Im experiencing with it now is the observability layer I have implemented inside it;

The whole stack is fairly simple Sveltekit + shadcn + python backend support;

For people that dont know: observability is a wide topic that encompasses all the ways you can trace the behavior and the way a software is running; usually to catch bugs before they become a problem for many users;

Im using sentry.io with the python sdk; since the most sensible part is the process where the backend goes to ask google for the youtube video and gets it transcription; many things can go wrong there;

Even by using known libraries you still get confronted with un-chaught exceptions edge cases and dead ends.

So my main use case for observability is to improve it on the go; all the super mega software engineers that want to dunk on me please do it, this is just a random project nobody pays me for so I do it my way, so I believe its perfectly normal to ship not a 100% tested software that I can improve over time;

The whole project has 0 integration and unit tests;

for the curious: https://pelicantools.app/tools/youtube-to-transcript

Comments URL: https://news.ycombinator.com/item?id=48401003

Points: 1

# Comments: 0

Categories: Hacker News

AI Leaders Call for Rules on Synthetic DNA to Limit Bioweapons Risk

CNET Feed - 6 hours 17 min ago
The heads of OpenAI, Anthropic, Google DeepMind and more signed an open letter raising concerns about the rising biosecurity risks posed by better AI models.
Categories: CNET

Belkin's Pricey New Battery Grip Makes My Switch 2 Feel Like a Steam Deck

CNET Feed - 6 hours 35 min ago
The $100 accessory includes a snap-on battery and large grips, but it makes the Switch 2 a bulky boy.
Categories: CNET

Offroad Emerges From Stealth With $7 Million to Tackle Enterprise Identity Risk

Security Week - 7 hours 31 min ago

As AI agents, machine identities, and third-party applications multiply across enterprises, Offroad is betting autonomous security agents can restore control over an increasingly unmanageable identity landscape.

The post Offroad Emerges From Stealth With $7 Million to Tackle Enterprise Identity Risk appeared first on SecurityWeek.

Categories: SecurityWeek

The UK’s competition watchdog has ruled that Google must provide online publishers and news organisations with the ability to opt out of their work being summarised by artificial intelligence, or otherwise used to train the company’s models

Computer Weekly Feed - 7 hours 55 min ago
The UK’s competition watchdog has ruled that Google must provide online publishers and news organisations with the ability to opt out of their work being summarised by artificial intelligence, or otherwise used to train the company’s models
Categories: Computer Weekly

Pages