Here is how to enable a 1M-token context window in Codex for GPT-5.6 Sol.
Even though we have tuned the context limit in Codex to be set optimally when it comes to performance and cost, this is a common ask, so here it is documented.
A larger context window lets Codex retain more code, tool output, and conversation history before summarizing older material. You need a model that supports it. And GPT-5.6 Sol, for example, has a documented 1,050,000-token window.
Open ~/.codex/config.toml and add or update these settings at the top level, before any [section] headers:
```
model = "gpt-5.6-sol"
model_context_window = 1000000
model_auto_compact_token_limit = 900000
```
The first setting selects the model. The second tells Codex to use a one-million-token context budget. The third starts automatic history compaction around 900,000 tokens, leaving some headroom. Restart Codex client and start a new session after saving.
To try the configuration for a single CLI session without changing your defaults:
```
codex -m gpt-5.6-sol \
-c model_context_window=1000000 \
-c model_auto_compact_token_limit=900000
```
Have fun, but also know that we tuned the default carefully!
Transparent backgrounds are now available in preview for GPT-Image-2 in the API.
Generate reusable assets you can place on any background—for product imagery, graphic design, website mockups, and marketing campaigns.
Welcome to Typefully
Here's what you can do with it:
📝 Write without distractions
⏰ Schedule your posts
📊 Track what resonates with analytics
🚀 Publish to X, LinkedIn, Threads and more
Typefully also helps you with AI ✨
Press ⌘ J on Mac (ctrl J on Windows) to try it.
Or click the "Refine" button under your posts to improve them.
Even better, visit Settings → AI & Agents to start using Typefully with your favorite AI apps and agents.
si entendés este flujo, ya entendés gran parte de cómo trabaja Claude Code.
en una tarea típica, el agente hace algo como esto:
1) entiende el contexto
> lee CLAUDE.md, usa la memoria disponible y analiza la conversación para entender el proyecto.
2) planifica
> arma una estrategia antes de empezar.
3) ejecuta
> lee archivos, edita código, ejecuta comandos y usa las tools necesarias.
durante la ejecución también toma decisiones:
- skill: cuando la tarea requiere un workflow o conocimiento reutilizable.
- MCP: cuando necesita usar tools de sistemas externos (por ej: abrir un PR en github).
- subagente: cuando conviene dividir la tarea o trabajar en paralelo.
4) evalúa
> analiza la salida de las tools, errores y tests.
> decide si ya terminó o si necesita seguir iterando.
> este ciclo de (ejecutar → evaluar) puede repetirse varias veces hasta completar la tarea.
5) aplica guardrails
> pide permisos cuando corresponde.
> ejecuta hooks automáticamente. por ejemplo, correr tests, validar reglas del proyecto o bloquear un commit si detecta información sensible.
6) administra el contexto
> usa compact cuando el contexto empieza a crecer demasiado.
Claude Code se entiende mucho mejor cuando dejás de pensar en features individuales y empezás a mirar el sistema completo.
We just open-sourced 50 skills that teach AI coding agents to make motion graphics.
Kinetic typography, data charts, explainers, TikTok/Reels, WebGL, Manim — 14 installable packs. Works with Claude Code, Cursor, Codex.
The problem: motion is the hardest thing to get right from a prompt. Timing, easing, safe areas, loops that actually loop. These skills encode how a professional motion designer would build it — with a verify loop so the agent checks its own output.
MIT licensed.
https://t.co/AoiJ51Fqh0
Claude Code creator:
"100% of our pull requests at Anrtopic are run by Claude Code. 80–90% of code review too.
The feature I’m using the most today is /loops. I’m not prompting Claude anymore - I’m building loops"
in this interview he breaks down exactly how a system changes everything:
- the memory and context features that turn Claude into a second brain
- the knowledge architecture most users don't know Claude can build
- the integration layer that connects Claude to your actual workflow
- why typing one question at a time is the most expensive way to use Claude
if you've been using Claude for months and still start every conversation from scratch with zero context, you don't have a Claude problem. you have a system problem
instead of another show tonight, watch this
make sure to bookmark it before it gets lost in your feed
full guide in the article below
@capcutapp CapCut me recuerda mis primeros pasos para implementar Videos en mis proyectos Web, para mis clientes y siempre era posible hacerlo. Sigan creciendo y éxito siempre.
The canvas is ready. What will you CRE[AI]TE?
We are excited to launch CapCut’s 1st AI Festival, a global celebration of original work and the imagination AI makes possible.
Submissions are now open through August 10, 2026.
From Films and Series to Creative and Commercial, CRE[AI]TE celebrates storytelling and creativity in all its forms.
Four tracks. One festival. Unlimited possibilities.
$200,000 in total cash awards, led by a Grand Prix, Category Winners, Category Honorees, the Student Vision Award, and Spotlight Selections. Plus film festival premieres, CapCut creator event screenings, industry connections, and more.
Everyone can apply for a free starter pack of CapCut credits and Pro access to kickstart your journey.
Submit your entries and learn more: https://t.co/4S2CTXFMwl
RT + comment in 9hr to get 200 free credits.
el creador de shadcn acaba de crear un skill para que puedas aprovechar Fable al máximo, ya que solo lo tendremos por otros pocos días 🔥
Se llama /improve y es una locura:
Usa Claude Fable para pensar como un senior dev (audita tu código, detecta problemas y genera planes brutales) y luego ejecutas todo con modelos más baratos y rápidos.
Puedes instalarlo con este comando:
npx skills add shadcn/improve
o puedes saber mas aquí:
https://t.co/1vmH6f2S3J