mxstbr

I run the team at OpenAI that's responsible for the ChatGPT App Store, Codex plugins, and all things MCP.

The thing that all these "MCP is dead" posts are missing is that whether or not MCP is used as a transport protocol is actually completely irrelevant.

The reason MCP isn't dead is because practically ~every company on the planet is building an MCP server. I know this because we interact with all of them. Most of these companies don't have a CLI. Many of these companies don't even have an external API! And yet, they're all building MCP servers.

And that's why MCP is not only not dead, but more important than ever.

Maybe we will turn every MCP server into a CLI under the hood. Maybe we'll use code mode. Maybe we'll implement tool search.

All of those are just implementation details to the much more important point: our AI agents are getting access to services they otherwise would never have had access to.[0] That's what matters.

So, is MCP dead as a direct communication layer for models to speak to? Maybe, maybe not. Is MCP dead as a protocol? Hell no, couldn't be further from the truth.

[0]: Although I will say the Codex app's computer & browser use features have made this statement a lot weaker than it used to be. If you haven't tried them yet—they're mindblowing.

https://x.com/XEng/status/1965226798460887127

Initial Show HN four years ago: https://news.ycombinator.com/item?id=26629429

Congrats to the Statsig team!

Initial Show HN four years ago: https://news.ycombinator.com/item?id=26629429

Congrats to the Statsig team!

It's interesting that he signs off with "So long, and thanks for all the fish," which is a quote from The Hitchhiker’s Guide to the Galaxy:

Just before the Vogons demolish Earth to make way for a hyperspace bypass, the dolphins, knowing what’s coming, leave the planet and their farewell message is “So long, and thanks for all the fish.”

I wonder what he's implying about GitHub…

I've been doing this for years with my meetings and I wish Google Calendar had it built in. I have to keep manually adjusting start times and it's a pain.

Oh man, I got excited to use this with our kids, but then saw that it requires a " mobile device management (MDM) solution and Managed Apple IDs." I hope they consider the family use case in the future.

Stellate CEO here; let me know if you have any questions about this!

Patrick Collison posted a summary of the results on Twitter in case you don't want to read the whole report: https://x.com/patrickc/status/1833127988512305352

TL;DR: "Mario Draghi's new report on EU competitiveness doesn't mince words."

Love seeing more innovation in this space! I know Oleg (former Webflow eng) has been working on https://webstudio.is for a while; also open-source and visual editing for React apps. How does Onlook compare?

I was thinking about changing my personal website's font to a monospaced one.

Anybody know which ones are particularly good for long-form text readability?

Bonus points if it's on Google Fonts.

I've been having a lot of fun building a digital garden for my WIP notes into my personal website: https://mxstbr.com/notes/digital-garden

My quests (goals) with this digital garden are:

1. Publish more than I did when I just had space for essays, which hopefully leads to…

2. Getting more input from people on my ideas

3. And have fun futzing with my digital garden technically

So far, so good!

I recently saw this related article[0] on the HN frontpage and I've already been applying it.

TL;DR: To build more quickly, do "outline speedrunning": Recursively outline an MVP, speedrun filling it in, and only then go back and perfect.

[0]: with me: https://learnhowtolearn.org/how-to-build-extremely-quickly/

Absolutely, if you're building a GraphQL server, parsing is almost certainly not your bottleneck. Stellate is effectively a GraphQL CDN, so for us, it is.

Our request flow effectively boils down to:

1. Parsing a GraphQL query to assess which types and fields are within it

2. Matching the configured cache rules to those types and fields

3. Caching partial results accordingly

So, for our specific use case, parsing is one of the main things our edge gateway does, and reducing overhead there allowed us to significantly improve our overall performance.

There is no source; black text on a white background. How do we know this is real?

Mark Larah at Yelp recently published their internal presentation on "Why GraphQL." Source: https://twitter.com/mark_larah/status/1777456733666840615

With three friends, we built and launched a SaaS product over a weekend and live-streamed ourselves doing it[0]: https://feedback.fish

It was a ton of fun, about 1,000 people tuned in over the course of those ~60 hours. The product was acquired about a year later.

[0] https://productweekend.live

Default GQL encoding is redundant and we care a lot about limiting size because many of our customers have low bandwidth

I'd love to know how you ended up combating this! I'm assuming something like graphql-crunch[0] or graphql-deduplicator[1], but I'd love to know what worked well in practice.

[0] https://github.com/banterfm/graphql-crunch

[1] https://github.com/gajus/graphql-deduplicator

I feel like this title is misleading compared to the original article. (cc @dang) Fly.io without cold starts (which is a one-line configuration change) is 2x faster than Cloudflare Workers.

That talk is the most mind blowing thing I've watched in years. Bonkers!

Here is a direct link to the repo: https://github.com/zed-industries/zed

Other way around! Notion bought Cron, which was around for a long time before Amie and (one could argue) inspired Amie.

See his recent tweet: https://x.com/paulg/status/1746341562286539118?s=46&t=hJLz5n...

This is the first German word I taught my (American) partner and it is, to this day, likely our most used German word. Such a core concept to our entire culture!

I don't get what this is. Is it a CMS? Or a database? Or both?

Personally, through all these conversations over the last years, it's become obvious to me that there is a better way for companies to manage the interface between frontend and backend that few of them know about, data layers.

We have the passion and the expertise to try and help a few more companies to use data layers and can justify it as a business because our main product at Stellate is GraphQl edge caching—which works with any GraphQL API, including those created by Fuse.js.

This is similar to why Vercel can invest so much in React and Next.js

Agreed. We initially used Timescale for our GraphQL Metrics product[0] but very quickly ran into scaling & performance issues. We switched to Clickhouse and have scaled 10.000x+ since with almost no issues.

[0] https://stellate.co/graphql-metrics

I want this taskbar app! (ideally with streaming transcription vs. at the end) How can I get it?

This person businesses.

Raycast Pro[0] has this + much more built-in for $8/mo. I have Capslock + A assigned as the hotkey and have superfast AI at my fingertips. 11/10 would recommend!

(on top of Raycast being a better Alfred/Spotlight anyway)

[0] https://www.raycast.com/pro