Luis Colunga

@0xSero I like to use the Librarian to look at new repositories and get a quick overview. You can just point it to the URL without cloning it. Prompt - Use Librarian to check {GITHUB URL} and provide me an overview

Hello Amp, Any tips from the pros?

@0xSero Not everyone gets them now. It's grandfathered for old accounts that aren't inactive.

Does everyone get a free hour of Amp per day? I am seeing 10$ of usage a day on mine which is around 1.5 hours in deep mode (GPT-5.*) or 45 mins in smart mode (Opus-4.*)

@LuisGBringas Funciona con VOLTE en las 3 redes?

Es bastante bueno, quieren speedtest?

@lcamtuf @tqbf @XorNinja A couple of days ago

THIS IS GOING JUST GREAT.

@jdrhyne Thanks for sharing this. Tried it on edge-case PDFs that usually break parsers (bad structure, messy tables), and it held up really well. Table extraction to Markdown was especially impressive. Not just context savings, the output quality is strong.

One of the things not talked enough about is context budget. Our document extraction reduces LLM context consumption by 92% compared to raw PDF processing. With our PDF-to-Markdown converter we're using 6x to 12x less context depending on the parser you're using. That leaves you and your Agent more tokens for reasoning and less wasted on data extraction.

@benvargas yep, that setup works. heads up though, if you're running Kimi-K2.5-Turbo via Fireworks, the generic chat-completions layer was dropping thinking tokens in my tests. switching to Anthropic directly fixed it and it feels a lot closer to opencode. luiscolunga.com/posts/factory-…

Don't actually need the legacy config.json that this article referenced... and vibeproxy uses CLIProxyAPI so for more advanced mac users, CLIProxyAPI can just be used too... Sample of my real settings.json with API Keys removed... gist.github.com/ben-vargas/070…

@0xSero Awesome

Giving away 5 Opencode Go subs Winners selected randomly from comments in 24 hours.

@0xSero Glad I wasn’t crazy. I hit the same issue with GPT-5.3-Codex via BYOK. Tool calls would drop and everything just stopped. Also had leftover legacy settings in settings.json that made things worse.

In the last 3 weeks: 1. They've updated the desktop app: - Project folders - Session search - Better support for BYOK - Skills viewer + their skills are top notch 2. BYOK has gotten much better - missions now accessible - you don't ever have to use any of their models - spec mode is phenomenal - Claude Opus can now do 20 hour runs ------ What's still broken: 1. GPT-5.3-Codex via BYOK drops tool calls and just stops working 2. Lots of flickering in Zed Overall happy customer. x.com/0xSero/status/…

@SowingAlphaSeed @the_green_lark I just remembered RPAR ETF implements the commodities portion of their portfolio with equities only. I wonder if you get similar beta numbers.

@the_green_lark Dang. Much higher than I was expecting.

Seems like a basket of these types of companies might be the best option for the commodity portion of an all-weather portfolio.

@SowingAlphaSeed What’s the IBKR model line?

Fortunately, that about bottom ticked it for the day. Still not a great day.

Ugly day!

@kai_brokering @claudeai @doodlestein flagging this in case you might be able to help connect those credits to a worthwhile endeavor

completely forgot i have $30k worth of @claudeai credits that expires in 3 weeks wtf am i supposed to do with all this😭

@antirez Did you ever try running tgterm as a macOS service (LaunchAgent plist), or do you prefer running it manually in the foreground?

Welcome to TgTerm: run your coding agents via Telegram for fun and profit: github.com/antirez/tgterm

@SowingAlphaSeed This is crazy. Any idea if the pair survived with borrow rates and availability factored in? Even at smaller sizes?

One AI ETF vs another

@SowingAlphaSeed 😂

Not quite X algo...

@khemaridh I wasn’t aware of it until now, but at a glance it looks very similar to Google’s Jules.

Doomscrolling on my phone is forever changed. (I set up Claude Code for Web.)

@vxunderground @ddd1ms Nice! I am in.

It is time for our first giveaway. We're giving away a Librem 14 from Purism. It's a fancy expensive $1,400+- laptop. Requirements: - Follow @ddd1ms on Xitter - Comment below Librem is a pro-privacy laptop that unironically comes with a fuckin' kill switches for mic, bluetooth, camera. It has Intel Management engine disabled. It runs PureOS, with app sandboxing, adblocking, tracking protection, etc. This laptop is basically a privacy nerd laptop. It also comes with a bunch of NSA stickers, HOPE (Hackers on Planet Earth) stickers, FBI Most Wanted stickers, etc. I forgot to ask for the specs on the laptop, but I'll get that stuff later on. Attached image is the laptop he'll mail to your home.

@EricBalchunas DOGE treasury HQ: Omaha, NE. Only 20 mins from Berkshire 🤣

CleanCore Solutions $ZONE is converting to become the first ever Dogecoin Treasury company in partnership with the House of Doge. Stock immediately plummets 59%. What a world.

@sinpozx Muy buen diseño. También qué bien que sean dos tarjetas físicas, y no un híbrido estilo la 2NOW, que usa el mismo EMV con dos aplicaciones.

@sinpozx En Telcel tienes plan o prepago actualmente?

@sinpozx Suena muy bien, ojalá puedas seguir compartiendo. La línea versus las otras que ya tenías fue razonable?