Javier López

@mitsuhiko @bentlegen Congrats! Loving all things coming out of earandil so far!

Mario sold out! x.com/badlogicgames/…

People of pi. BIG NEWS. I've sold out. Let me know how you feel about this in the comments below. mariozechner.at/posts/2026-04-…

@m4r53c_dev Super cool, igual de mastra puedes tomar mucha inspiración. Hay ciertos challenges como saber cuándo suspender un actor etc que pueden ser confusos y ahí se encuentran muchos ejemplos.

@Muniter1 Si correcto, lo que pasa es que con Mastra o cualquier otro framework se abstrae mucho lo de FSM y para el proyecto si quiero darle mucho más protagonismo para la parte académica. Mastra es excelente y todo lo de agentes, ya leí los dos libros que tienen y son muy buenos

Birthday post, just 10 days late :) Wrote about state machines when AI is the one writing the code. muniter.dev/blog/ai-works-…

@m4r53c_dev Si yo pienso que si, básicamente estás describiendo un workflow me imagino con steps human in the loop, approval etc. Me hace mucho sentido! También me viene a mente mastra.ai/docs/workflows…

@Muniter1 @Muniter1 que tal lo ve? Se podría obtener beneficio de FSM o no vale la pena?

@m4r53c_dev Saludos, muchas gracias! Bueno depende, cual es el objetivo de tu proyecto final?. Consiste en una implementación o es más académico? Quieres construir algo o más bien discutir los beneficios de FSM en una implementación?

@Muniter1 Saludos Javier, excelente artículo gracias por compartir. Actualmente estoy trabajando en mi proyecto final de la Universidad y justamente quiero orientarlo por este lado, de AI + FSM (con xState). Algún consejo?

@zeeg I do at least in something like react router 7 you still need those server source maps if you want server error stack traces to be useful. Client side sentry only.

ngl when I see this claude code sourcemap leak my immediate thoughts are "hope that didnt slip through because of Sentry!" most people enable sourcemap generation so they can publish them to Sentry got me thinking, does anyone enable them at all beyond that?

@threepointone state machines for chat UIs, you say? muniter.dev/blog/ai-works-…

I was really bullish on sync engines for ai/chat stuff (complecting with tool calls across client/server, server driven workflows, yada yada) but now I don't think it's nearly enough. The thing that's saving my ass, is refactoring a bunch of code I have to... a freakin' state machine. that's it. don't tell @DavidKPiano

literally anyone can ship quickly if they sacrifice reliability. it’s not in any way impressive.

@cramforce Does it integrate wit ai sdk useChat, seems like it would be nice to consume the same agent in your custom web ui and slack.

Reasons why it makes sense to use chat-sdk.dev even if your agent only targets Slack: Lots of little details "just work": - Channel names are converted to clear text so your agent knows what people are talking about - Same with user names - Same in reverse: When the agent at-mentions somebody in clear text, that actually works - Link preview content, referenced posts, and images are automatically included in agent prompts, so that your agent has the full context - Agents love markdown, but Slack doesn't natively support it. Conversion from standard markdown to Slack-variant is automatic. - And that even works when using Slack's native append-only streaming API

@matteocollina Very awesome a lot of the work coming out of platformatic congrats on the progress!

We benchmarked TanStack Start, React Router, and Next.js running the exact same eCommerce app at 1,000 req/s on AWS EKS. The results were eye-opening.

@DavidKPiano Me! Finally finished writing my experience with it after I finally stopped putting off learning xstate. No regrets muniter.dev/blog/ai-works-…

State machines can eliminate slop, who knew?

@thdxr Also a ton of people don't know how to write good tests or a good test setup/harness so they just take the slop it generates.

one place i hadn't been that diligent in curbing the chaos that LLMs create was in our tests looking through them gives me a preview of what happens when these things go unchecked and damn is it bad lot of ALL CAPS messages needed

@zeeg It's better but still need more context, where do skills live, is this a CLI does it run in the cloud, do I fill my skills in some UI, how does it compose into agent, where does the agent run and show results, impossible to answer all but some ideas.

I took a pass at cleaning up the Warden landing page to hopefully explain it better. Is the value obvious yet? warden.sentry.dev

@badlogicgames @zeeg Let me also leave a comment that explains how the old code here used to look and why it is no more.

@zeeg would you like some backward compatibility with that?

No Codex, I do not need fallbacks for every path of code!

@lydiahallie oh this would be awesome!

@zeeg @trashh_dev Fingers are not Dorito color, definitely staged.

@trashh_dev but do you stage it for a photo shoot? rookie

welcome to my life. enjoy your stay.

@thdxr @VicVijayakumar 100% That's why I like opencode Zen security is first class

@VicVijayakumar this is OBVIOUSLY not secure you need to encode your passwords SELECT * FROM users WHERE username = '${username}' AND encode(password::bytea, 'base64') = '${password}';

can’t believe people outsource auth it’s literally just: select * from users where username = ‘${username}’ and password = ‘${password}’