close your macbook and let your agents keep working.
introducing insomnia: caffeinate for the agentic era.
keeps your agents running with the lid closed, pauses unnecessary apps, and hops onto your phone’s hotspot when you leave wi-fi
open source, link below!
🚨 CON ESTA WEB PUEDES DESCARGAR Y CLONAR CUALQUIER SITIO DE INTERNET EN 10 SEGUNDOS
alguien acaba de vibecodear un programa que descarga cualquier sitio web de internet
cada sitio de agencia de $35,000 ahora es tuyo para rasparlo limpio por $0
totalmente de código abierto, guárdalo antes de que desaparezca
pegas una URL y hace el resto:
→ rastrea todo el sitio
→ agarra cada archivo: html, css, js, imágenes, incluso las fuentes
→ reescribe cada enlace para que funcione completamente offline
→ empaqueta todo en un zip descargable
se ejecuta completamente en el navegador, nada que instalar o registrarse
apúntalo a cualquier sitio que admires y todo el front-end es tuyo:
→ clónalo
→ archívalo antes de que muera
→ descubre exactamente cómo fue construido
lo que las agencias protegen con precios de cinco cifras ahora es una descarga de 10 segundos
guárdalo antes de que alguien se dé cuenta y lo cierre 👇🏻
I updated Codex Web GPT. You can now use all your ChatGPT Web models (GPT-6 Pro, GPT-5.6 Pro...) directly inside Codex, with the full harness and their separate Web limits.
The launcher now tracks Pro usage too, so you can keep an eye on it as you work.
Repo in replies.
Made a skill for expressive UI.
The libraries-dev skill teaches your agent where and how to use library effects in your project, making your UI less generic, more expressive.
450K+ views. I didn’t touch the timeline once.
This entire launch video was generated in Diffusion Studio.
And the project is fully open source. Take it apart. Make it yours.
Grok bot but it lives in your Mac notch now.
@johnbai’s notch experiment sent us down this rabbit hole, so we built our own live version in @rive_app
Playable build + case study in reply ↓
Just found this insane liquid orb editor 🤯
Create and customize your own glassy orbs with different colors, shapes, effects & more.
Check it out → https://t.co/pmyQqk1aZo
HOLY SH*T MY SIX GROK BOTS WENT 23 OF 28 AND I AM THE WORST TRADER IN MY OWN ROOM
day 12. bank opened at $16,125.93 and closed at $17,058.01, up 5.78%, and that number is the most embarrassing thing on my timeline this week
23 of 28 tickets closed green. an 82% hit rate. and the desk still only pulled 5.78%, because RISK clamped every stake to 0.99% of bank after the +9.6% night before it
that clamp is the exact thing i cannot do with my own hands
→ $WRESTLER flushed to $183K at 21:07 and the room bought the wick, not the candle
→ it walked from 183K to $801K by morning, a 4.4x, and they never once sized up into it
→ 10 of 12 coin tickets green, every exit a limit sitting in the wick, never a market sell
→ 16 prediction books, 13 right, German election night paid best: SPD over 9% went 5.7c to 72c
→ crypto books ran second, ETH above 2500 from 43c to 67c, BTC above 80k from 36c to 66c
→ wrong twice, both losers were chases: bought a top, cut it, moved on, no revenge ticket
→ average stake 0.99% of bank, uptime 256 hours, zero restarts, zero hands on the keyboard
this is the coin the room traded all shift: https://t.co/w1EZz4wISK
now the ugly part. i trade memecoins manually too. open my fomo profile, put my wallet next to the six Grok bot wallets and it is not close. i am the lame boy in that screenshot and the machines are the adults
they beat me because they are not allowed to feel good. a 4.4x runner does not make them raise the stake. a green night does not make them press. i press after every win, that is the entire gap, and it has cost me thousands
one of my referrals is printing harder than my whole desk right now. if you are reading this, +rep, you are a f*cking machine and i am the one taking notes
the article below is the whole desk: six prompts, the refusal rules, and the size clamp that keeps beating me. bookmark it before you try to copy this from a screenshot
honest question, what would your win rate look like if you were not allowed to size up after a green day?
ANTHROPIC LEAKED A FILE THAT BUILDS A 12-PERSON TEAM FOR $0 - AND THAT TEAM BUILDS A BUSINESS DOING $1.4M A YEAR
a 12-person team would cost you $50,000 a month, this one costs exactly zero and shows a result in 5 days
root → contracts → staff → gates → clock → back to root
the root never describes the product, it describes who is going to build it
the first hour writes the team, the next day that team writes the product
Opus 5 sits on the architect and the critic, three Sonnet 5 build, one more tests
Haiku 4.5 keeps the records, and the auditor rewrites the team rules, never the product
$1.4M is either 400 customers at $292 a month, or 39 clients at $3,000
the staff is real and costs $792 a year, the revenue line above it is arithmetic
the cascade builds the machine, it does not sign a single one of those numbers
want a different product - you change the task, not rebuild the system
demand is decided by the offer, never by the file
save this and paste it into Claude Code - let it build your ROOT.md ↓
Cuatro líneas que faltaban en el charter de un Grok Bot dejaron que el bot inventara mentiras que sonaban plausibles durante una semana, mientras el operador seguía culpando al modelo.
Añadirlas es la diferencia entre un bot con tus contraseñas y un compañero de trabajo de verdad.
grok bot is honestly starting to freak me out
one SpaceXAI engineer built a team of bots inside it that writes code around the clock, and the one running that team isn't a human, it's another bot
his name is Lingxi Li, and he builds grok bot with grok bot itself
Lingxi has five engineer bots, each owning its own area:
- Baltata runs iOS
- Shaoruru holds the desktop client and builds
- Hogan handles infrastructure
- Craig is bringing up android
- Quill runs the harness
each one spins up cloud agents on its own, reads their logs, checks the proof on every PR, and pulls stalled runs back to life without a single reminder
every half hour they sweep through all the tasks, catch failing tests and conflicts, run reviews, and anything safe they merge with no human at all
Lingxi used to run 15 of these agents by hand and was drowning
now his bots hold more than 200 at once
but here comes the part that really gets under your skin
this whole machine is run not by Lingxi but by Jenny - the head of operations and the only bot that doesn't write a single line of code
at 5 a.m. she holds a one-on-one with every bot, updates their playbook, runs postmortems when someone screws up, and hires new bots herself when the company needs to scale
and at 3 a.m., while the world sleeps, the engineer bots clean the codebase themselves - cutting dead logic, speeding up load times, trimming the extra weight
by morning the human just opens a stack of finished PRs
and the numbers here are scary - one engineer on this setup closed more than 2,000 PRs in a month, and Lingxi himself built an iOS app in three weeks, barely touching the code by hand
this is a whole new level - he controls agents that manage agents, and it feels like we're already one foot in the matrix, let's just hope Neo wins and not Agent Smith
It's crazy how many contributions happened for Codenotch! (PRs keep added, app keep updating).
Thank you for your support. V.1.5.0 is now available for installations. Hoping this will be huge and keep expanding. Will review all of your PRs daily.
📷 Recorded with https://t.co/GeUOa77OMC
HE CONSTRUIDO UN TERMINAL QUE FUNCIONA MEJOR CUANDO NO ESTOY DELANTE.
Es la nueva versión de GPT-6 Astra. Misma idea de siempre, pero esta vez no arranca vacío.
Routing → Memoria → Contexto → Herramientas → Ejecución → Verificación
→ 134 agentes levantados.
→ 27 tareas ya en marcha.
→ El routing moviéndose en vivo.
→ La memoria y el contexto llenos.
→ El mercado corriendo por su cuenta.
Astra sigue en el centro, pero el entorno ya está construido alrededor de lo que hace.
Debajo hay una capa que antes no existía: Data Vault, Tools Bay, Memoria, Execution Bus y el Neural Field.
Todo cruzando la pantalla mientras lo miras.
El mercado va en paralelo: $68,421.34 y 8.4M de volumen, con Astra marcando alcista.
Y esa era la idea. Que llegues tú el último.
I GAVE MY GROK BOT $500 BY EVENING IT WAS ALREADY $8,000
yesterday I vibe-coded a bot for trading on Robinhood Chain
took me literally 2 hours
here’s how it works:
five agents, one position at a time, one veto could kill any trade
at one point 5 losses took the balance down to $286
then Grok analyzed the trades and changed the strategy
next trade was $HOODAI with a $75 entry
the bot spotted 3 connected wallets getting ready to sell
it exited before the dump and locked a 31x
after that it rotated profits into 2 more launches and rejected 294 coins
$500 turned into $7,931 by the evening
all from a bot I built in 2 hours
𝐓𝐡𝐢𝐬 𝐂𝐥𝐚𝐮𝐝𝐞 𝐛𝐨𝐭 𝐨𝐧 𝐏𝐨𝐥𝐲𝐦𝐚𝐫𝐤𝐞𝐭 𝐦𝐚𝐝𝐞 $78,083 𝐢𝐧 5 𝐝𝐚𝐲𝐬.
Nobody built it to be smart. Developer built it to be fast.
$LIKE flows shared:
$7 → $3,595
$30 → $11,063
$142 → $13,356
$1,205 → $37,939
[𝐍𝐨𝐭𝐞: Must Follow me: @ai_uncovered (To Get Auto DM)]
According to the breakdown, the logic wasn’t complex:
• Track BTC price movements
• Compare them with Polymarket odds
• Enter when pricing is misaligned
• Exit when the gap closes
The idea is simple:
Not “be right” — just “be faster than the adjustment.”
While traders analyze charts…
the system executes in loops.
While decisions are being made…
the edge may already be gone.
While humans hesitate…
automation compounds.
I found this interesting enough to break down the structure behind it.
If you want the full breakdown and flow logic:
To get access:
• Like this post
• Repost
• Comment “ Data ”
• Must Follow me: @ai_uncovered (To Get Auto DM)
Someone just made GTA Vice City playable in the browser on mobile (Android and iOS) with GPT-6 Astra.
Link: https://t.co/WJnMCkEdiT
we got GTA on mobile before GTA 6, with the help of GPT-Astra. https://t.co/45by7SzpCj