Me flipa cómo mi stack cambia últimamente cada pocas semanas, así que dejo este post fijado y lo iré actualizando cada vez que cambie algo 👾
Mayo 2026:
Terminal: Cmux <@manaflowai>
Skills: GitHub skills de <@mattpocockuk>
Agent Runtime: Pi coding agent <@badlogicgames>
Local LLM: Qwen 3.6 35B <@Alibaba_Qwen>
Frontier LLM: GPT 5.5 en medium <@OpenAI>
Introducing Flue — The First Agent Harness Framework
Flue is a TypeScript framework for building the next generation of agents, designed around a built-in agent harness.
Flue is like Claude Code, but 100% headless and programmable. There's no baked in assumption like requiring a human operator to function. No TUI. No GUI. Just TypeScript.
But using Flue feels like using Claude Code. The agents you build act autonomously to solve problems and complete tasks. They require very little code to run. Most of the "logic" lives in Markdown: skills and context and AGENTS.md.
Flue is like Astro or Next.js for agents (not surprising, given my background 🙃). It's not another AI SDK. It's a proper runtime-agnostic framework. Write once, build, and deploy your agents anywhere (Node.js, Cloudflare, GitHub Actions, GitLab CI/CD, etc).
We originally built Flue to power AI workflows inside of the Astro GitHub repo. But then @_bgiori got his hands on it, and we realized that every agent needs a framework like Flue, not just us.
Check it out! It's early, but I'm curious to hear what people think. Are agents ready for their library -> framework moment?
We didn't ship HDR and ask if the output held up. We asked studios to break the tech.
Gear Productions. Magnopus. Asteria.
The result: an HDR IC-LoRA on LTX-2.3 producing 16-bit HDR EXR files.
Gear ships LTX shots as final deliverables. Magnopus integrates AI assets onto LED walls. Asteria used the workflow on ChikaBOOM!, a Tribeca 2026 selection.
The line between generated and filmed disappears. Their words, not ours.
Check out the full story.
https://t.co/9d8uPE8a90
Pi coding agent de @badlogicgames + "local" qwen 3.6 35B de @barckcode + un mínimo de skills (destaco las de @mattpocockuk) para mi codebase en DDD y oye, va de cine. El modelo se nota que está un poco por detrás de los frontier pero considerando la diferencia de precio... más que apañado voy.
turns out not killing the prefix cache all the time and notnhaving a humongous set of tools and a massive system prompt is good for local model use.
who'd have thunk.
https://t.co/lWjRoikJeM
7 días de viaje familiar perfecto para desconectar y mi cabeza no para de recordarme todas las nuevas mierdas de IA que están saliendo y me estoy perdiendo 🤡
Hace solo un par de semanas que publiqué este post a modo de "idea loca" y bueno..
Hoy ya acabo de solicitar el segundo server para la comunidad de NaN:
https://t.co/WFOQPdxSet
Y encima todo el movimiento que esta teniendo y la waitlist que no para de crecer, enserio, gracias♥️