Angehefteter Tweet
Rhys
10.4K posts
Rhys
@RhysSullivan
Working @opencode | personal projs google index discord - https://t.co/gaRhpx9lkW | let agents call apis https://t.co/gIFOwRUrLU
San Francisco, CA Beigetreten Kasım 2007
1.2K Folgt53.2K Follower
@hichaelmart @swalker326 @felixrieseberg @vite_js @tan_stack It might have been but the second part implies that it's faster than a properly built SSR (technically PPR) setup when it wouldn't be
>Time to first byte is down 65% at p75, prompts show up 50% sooner
This is mainly a meaningless line as it's not an actual comparison
English
@RhysSullivan @swalker326 @felixrieseberg @vite_js @tan_stack "We moved our architecture from SSR to a static setup".
"It's faster".
Never said they moved *to* make it faster. May very well have been for simplicity?
English
A small ship I love: We made Claude.ai and our desktop apps meaningful faster this week.
We moved our architecture from SSR to a static @vite_js & @tan_stack router setup that we can serve straight from workers at the edge. Time to first byte is down 65% at p75, prompts show up 50% sooner, navigation is snappier.
We're not done (not even close!) but we care and we'll keep chipping away. Aiming to make Claude a little better every day.
English
@hichaelmart @swalker326 @felixrieseberg @vite_js @tan_stack That’s again not the point of the tweet
NextJS is not easy to use properly and I have a lot of pain points with it but the original post about their site being faster because they abandoned SSR is just incorrect, it would be more accurate to say they moved for the simplicity
English
@RhysSullivan @swalker326 @felixrieseberg @vite_js @tan_stack "if used properly" is doing a lot of work here though. The surface area of PPR vs an SPA is much larger. Far more things can go wrong. Far more things need to be done just right. Is the extra complexity worth the payoff? Sometimes! Sometimes not.
English
@noobpsyborg42 @tszzl My read of the tweet is unrelated to MCP, it’s moreso for platforms that have value in them but refuse to be accessible to agents
An example of this is I want my agent to order groceries weekly from Amazon but can’t do that automation since they block it
English
@tszzl I’m not hugely concerned on this one, either companies provide nice ways to let agents access their service, or people work around it with automation and headless browsers
The pain and security issues of the latter will likely force the former to be common place
English
@maxleiter github.com/vercel/vercel-…
skills become MCP resources
commands are just MCP tools
agents are MCP prompts
npx vercel-mcp@latest, use stdio for the MCP transport - gives you a typed auto updating version of the vercel cli
English
@BrandonWaselnuk customers wouldnt edit the files same way they can't edit docs on a companies website today
English
@RhysSullivan Go on 👀👀
How are you handling updating a skill without overriding if a customer edited the file?
Just saying if it’s in your namespace you might update it? Like ‘company-skill’?
English
@morganlinton token usage is a factor of the execution environment not MCP, a good implementation of MCP (on the caller agent end) is likely more token efficient but that's just starting to happen
English
@RhysSullivan Yup, working well, and really liking the token usage vs. MCP, so much more efficient.
English
can't believe i just realized the ball guys' last name is Dickens
Ben Dicken@BenjDicken
Glad OpenAI bought the fast Python package manager.
English
@swalker326 @felixrieseberg @vite_js @tan_stack what type of caching are you talking about, public CDN cache i.e public pages or private data that has to live in memory i.e ai chats
English
@RhysSullivan @felixrieseberg @vite_js @tan_stack I’d love to see an apples-to-apples benchmark. For mostly cacheable routes, TanStack with intent prefetch plus an edge-cached static shell can outperform SSR setups on TTFB and navigation feel. For highly personalized first-render content, streaming SSR may still win.
English
@swalker326 @felixrieseberg @vite_js @tan_stack i don't disagree but that's not the claim that's being made in the tweet
SSR (w/ loading shells) would probably make their site faster if used properly but it doesn't sound like they were using it correctly
English
@RhysSullivan @felixrieseberg @vite_js @tan_stack Vite + TanStack tends to provide better DX than Next (Turbo/Webpack), especially in local dev speed. server startup, HMR response and iteration speed.
English
@tobiastornros not at frontend so i've still got an anthropic sub for that but if i had to pick one it'd be codex
when i say its better its also maybe only like 10%, the models have both gotten very good but they still have their quirks
English
@RhysSullivan It should say when you are in thinking/pro right in the composer on web, no?
English
