Building @turborepo at @vercel | Used to play pro ⚾, now I’m a full-stack comedian.
Joined August 2014
- Tweets8.8K
- Following341
- Followers6.8K
- Likes17.2K
My CTO found my repo.
ALT Draft GitHub PR by cramforce from branch experimental-turbo-trace, showing +441,926 additions and -151 deletions.
Anthony Shew retweeted
Vercel Security Review, less than 1 week since launch:
✓ 𝟱𝟳,𝟴𝟵𝟳 vulnerabilities identified
✓ 𝟲,𝟬𝟮𝟵 critical or high
✓ 𝟭𝟰,𝟲𝟲𝟱 already patched by Agent
I'm shocked at ① the number of critical vulnerabilities that would go unnoticed and ② the ease at which the agent automatically fixes them.
You can turn it on for any repo, no matter where you deploy to.
CI myths the Twitterverse is believing right now:
- "Larger or faster machines will solve my problems."
- "Deleting unit tests is a good thing to do."
- "Agentic test selection is the answer."
- "Coding agents don't need CI."
- "Monorepos don't work for agents."
Anthony Shew retweeted
small monday 🚢: @vercel CI status is now mirrored in your browser favicon so you can keep an eye on build progress at a glance
fx ask "fix typo" \
--model=anthropic/claude-fable-5.2 \
--effort=megamind \
--fast \
--ultrafast
13
846
Anthony Shew retweeted
Works without thinking (Agents set it all up without an issue)
Anthony Shew retweeted
libghostty 1.0 is here! I started Ghostty ~5yrs ago to work towards this. Today, libghostty is in hundreds of projects used by tens of millions of people. 1.0 marks ABI stability, quality, and an invite for distros to begin packaging. mitchellh.com/writing/libgho…
Since starting the Ghostty project in 2022, the most common question I've been asked is "why?". Today, I'll answer this by showing the broad industry impact libghostty has already had.
On top of JetBrains and Neovim already mentioned (that alone being over 10 million users):
-KDE developers are experimenting with a Konsole replacement built on it.
- Vercel's Turborepo uses it to render task output in its TUI.
- devenv uses it to drive shells.
- Antithesis used it to fuzz TUI programs and found real bugs in btop and others.
-Ghostel brings it to Emacs.
- ghostty-web is the WASM build with an Xterm.js-compatible API.
- Commercial products like OrbStack embed it.
- And a whole wave of new terminals and "agentic" coding tools, such as herdr and cmux, have been built on it rather than writing yet another emulator.
There are community bindings for Rust, Go, JavaScript, Swift, Python, .NET, Elixir, Dart, Java, and more, and the awesome-libghostty list tracks well over a hundred projects.
More people today use a terminal powered by libghostty than use Ghostty the application, by a very large multiple. And libghostty is having a much larger impact than being "just" a terminal emulator.
I'm excited to see where people take it from here.
I've been lying this whole time. I'm sorry, I just found out. I'm just as disappointed in me as you are.
That 2-6x faster number I've quoted so much...
After some more dogfooding, our results are showing the real numbers are...
2-10x faster. 😄
Introducing Vercel CI
• Intelligently skips tasks with Turborepo
• 2-6× faster CI in our internal repositories
• Optimized for agents via Vercel MCP and CLI
After nearly 8 million tasks, we're excited for the future of agentic CI. In beta on all plans.
vercel.com/blog/introducing-…
FYI: You don't have to deploy to Vercel to use Vercel CI.
Vercel CI now works on GitHub repos without a Vercel project.
Install the Vercel GitHub App and link your GitHub account to Vercel. Then run this in your repo root:
▲ ~/ vercel ci enable
vercel.com/changelog/vercel-…
This is why Vercel CI rounds to seconds.
We knew we were building a faster system, and we didn't want to penalize you for being fast.
ALT Vercel Docs: Vercel CI calculates CPU minutes as requested CPUs × run time, including shutdown. The highlighted text says time "rounds up to the next second," with a 1-second minimum.
CI pro tip:
GitHub Actions (and Blacksmith etc.) round every job UP to the nearest minute
... so a 10 second job still costs a full minute!
If you’ve got a bunch of short jobs, you’re potentially better off combining them into one and parallelizing within
4
73
4,329
"A faster system, you say?!"
Yes, a faster system, I say.
xitter.cf/anthonysheww/status/21…
This repo has 1,455 packages and 27,323 tasks in it.
It just completed CI for my PR in 1m26s.
Work smarter, not harder.
ALT CI run summary showing all 500 tasks completed in 1 minute 26 seconds, with a shocked-face emoji and arrows pointing to both numbers.
1
268
This repo has 1,455 packages and 27,323 tasks in it.
It just completed CI for my PR in 1m26s.
Work smarter, not harder.
ALT CI run summary showing all 500 tasks completed in 1 minute 26 seconds, with a shocked-face emoji and arrows pointing to both numbers.
Why is Vercel CI faster than just "faster CPUs"?
Caching, shared dependency installs, maximum concurrency.
1. Task caching by default
Tasks only run if they could have been impacted by the current diff. And it isn't like the task boots and then terminates because there is nothing to do: The task never even starts because the orchestrator is aware of the caching.
This saves wall-clock time and $$$.
2. It installs dependencies once and then shares them between all tasks 🤯
Have you ever babysat a CI job 30 minutes in and wondered why it is running `pnpm install` again? Vercel CI never does that.
This saves wall-clock time and $$$
3. It runs all tasks at maximum concurrency. In legacy CI systems you can manually configure task fan-out, but that is tedious and often bit rots. Vercel CI always runs tasks as soon as possible and all tasks that can be concurrent get their own VM and run in parallel.
Increasing concurrency is cost-neutral while massively reducing wall-clock time.
vercel.com/blog/introducing-…
Anthony Shew retweeted
Why is Vercel CI faster than just "faster CPUs"?
Caching, shared dependency installs, maximum concurrency.
1. Task caching by default
Tasks only run if they could have been impacted by the current diff. And it isn't like the task boots and then terminates because there is nothing to do: The task never even starts because the orchestrator is aware of the caching.
This saves wall-clock time and $$$.
2. It installs dependencies once and then shares them between all tasks 🤯
Have you ever babysat a CI job 30 minutes in and wondered why it is running `pnpm install` again? Vercel CI never does that.
This saves wall-clock time and $$$
3. It runs all tasks at maximum concurrency. In legacy CI systems you can manually configure task fan-out, but that is tedious and often bit rots. Vercel CI always runs tasks as soon as possible and all tasks that can be concurrent get their own VM and run in parallel.
Increasing concurrency is cost-neutral while massively reducing wall-clock time.
vercel.com/blog/introducing-…
It keeps happening.
ALT Bar chart: Vercel CI with 6 shards finished in 8m 14s. Unsharded GitHub Actions took 38m 10s on the same commit. Both passed all 2,451 tests.
"Hey, Anthony, how did Launch Day go?"
*gestures broadly at customer feedback*
![Slack thread screenshot. A customer posts: "Checks took 13.6 seconds. Total time was 37.2 seconds, including 23.7 seconds of Vercel startup. This is going to be crazy for velocity!" Anthony Shew replies: "Love this. Do you know what your timings were before?" The customer answers: "2m36s on average with [redacted] and before that 5m47s with [redacted] running with our self hosted runners in AWS. I have been on a bit of a crusade with it…" Anthony replies: "Absolutely stellar."](https://pbs.twimg.com/media/HVzokZKbgAAHR5i.jpg?name=orig)
ALT Slack thread screenshot. A customer posts: "Checks took 13.6 seconds. Total time was 37.2 seconds, including 23.7 seconds of Vercel startup. This is going to be crazy for velocity!" Anthony Shew replies: "Love this. Do you know what your timings were before?" The customer answers: "2m36s on average with [redacted] and before that 5m47s with [redacted] running with our self hosted runners in AWS. I have been on a bit of a crusade with it…" Anthony replies: "Absolutely stellar."
ALT Slack message screenshot from a customer: "I'm so glad to not have to focus on CI anymore 😂 You can have it"