A surprising conclusion I have come to from working with agents. blog.exe.dev/devtools-mus...
Devtools must be open source - exe.dev blog
We have entered the age of personalized software.
blog.exe.dev
David Crawshaw
@crawshaw.io
ceo exe.dev, tailscale co-founder, programmer
A surprising conclusion I have come to from working with agents. blog.exe.dev/devtools-mus...
Devtools must be open source - exe.dev blog
We have entered the age of personalized software.
blog.exe.dev
I told someone about @exe.dev today and told them not to google it. Open a terminal and just ssh. It's the 🤯 moment I love to see
The hardest part of developing little programs for yourself right now is setting up an environment to run them. This is the idea of @exe.dev, open the app on your phone, start typing, you have a program on a durable Linux sandbox.
Four quick prompts to make a custom feed bsky.app/profile/did:... for myself on an @exe.dev VM.
Billing turns out to be a fascinating async database replication problem.
"Instead of having billing API calls littered in the code, what if we issued billable facts about the state of something and reconciled billing once the new facts were settled?" blog.exe.dev/billable-facts
Happy @exe.dev-on-youtube day for all who celebrate. www.youtube.com/watch?v=BhBF...
Agents Ship from Anywhere: exe.dev on iOS
YouTube video by Official Channel for Exe Dev
youtube.com
PSA: You can now interact with your VM on Slack blog.exe.dev/run-a-slack-...
Run a Slack Bot from Your VM (Without Giving It the Keys) - exe.dev blog
blog.exe.dev
Built a distributed DNS server in ~a week. Rebuilt twice on purpose. Shipped -> went on vacation. DNS incidents a month later: 0. How we did it (hint: the answer isn't vibe-coding): blog.exe.dev/claude-is-no...
Claude Is Not a Compiler - exe.dev blog
blog.exe.dev
Setting up Gerrit on @exe.dev with X-ExeDev-Email auth took 60s! However, that doesn't make git work. I hacked together a little proxy that does a tiny amount of OAuth2, exchanging exe.dev auth in the browser for a token that works with git-credential-oauth.
GitHub - filippo-claude/gerrit-exedev-auth-proxy: Authenticate Gerrit web and Git smart HTTP with exe.dev OAuth
Authenticate Gerrit web and Git smart HTTP with exe.dev OAuth - filippo-claude/gerrit-exedev-auth-proxy
github.com
We're going to be sharing more about how people use exe to build cool stuff. First up—a conservationist who turned 40 terabytes of public data into a video game: blog.exe.dev/meet-the-con...
Meet the Conservationist Who Turned 40 Terabytes of Public Data Into a Video Game - exe.dev blog
Raffael Hickisch spent years dreaming up conservation tools he couldn't afford to build. Then he built them himself, with an AI agent and a country's worth of public data.
blog.exe.dev
New episode of @fafo.fm with @crawshaw.io about @exe.dev scheduled for Monday Have some more outages included in the newsletter. If you like reading those kinds of things subscribe at sub.fafo.fm
Get alerts in your inbox
We'll send you emails about new episodes, special content, swag, and live or in-person events.
sub.fafo.fm
I’ve been considering buying a new Mac mini or Mac Studio (yes, yes… I’m aware of the price increases) but after reading this I might just stick with my MacBook Air and just use @exe.dev
Building Software From My Phone - exe.dev blog
blog.exe.dev
Used exe.dev to build and host my first custom Bluesky feed: bsky.app/profile/will... so that I don't miss any great baseball goofs. I was worried it might use a good bit of my vcpu quota but consuming the jetstream is way less resource intensive than I would have guessed
exe.dev - ssh exe.dev
Build apps or SSH into a persistent Linux VM. ssh exe.dev.
exe.dev
All these "clientless" compute options appearing over SSH. Someone tried to sell me one today in an HN comment thread. Truly, the sincerest form of flattery.
I am convinced that commercial LLM offerings are bad at writing as a side effect of their post-training. I saw better writing from open base models 18 months ago. I cannot even trust a "frontier" model to write dev docs. More evidence we need a future of many models.
nyc fighting waymo is like reading about people who used to fight seatbelts. every time I see young children cross at the intersection near my house where a pedestrian was killed earlier this year, I wish autonomous vehicles were in my neighborhood.
This is a limit of large human institutions. A large group cannot trust its own people, because large groups include bad people. So you have to build in checks and enforce them. But some people cannot operate effectively under those constraints. So some of the best people stick to small teams.
To get good value out of agents you need a good development cycle. For an iOS app that means: while you are playing with the app, see a bug, you flip over to the agent, explain the problem, and then your app is updated with the fix. It's a REPL for iOS dev. Takes a lot of setup.
Behind the scenes on how the exe.dev iOS app was built: blog.exe.dev/ios-developm.... Step 1: buy mac mini. Step 2: give Shelley access to simulator. Step 3: over the air updates, Step 4: prompt app into existence from itself. Also, gratuitous profiler. boldsoftware.github.io/public_html/...
Me, reaping: *startled* *fart noises emanating from Mac* TEN MIN PRIOR Me, sowing: "Shelley, build an iPad sound machine; test with the simulator." THIRTY MIN PRIOR "I learned how to build iPad apps at work; kids, what shall we make?" TODAY exe.dev has an iOS app blog.exe.dev/ios
It makes me really happy to think of all the people who went into aerospace, not because they saw it as a path to wealth but as a way to do something that needed doing, also ending up with a financial windfall. They deserve the win.
New day, new integration: bring your ChatGPT subscription to exe. Power Shelley with it for all your VMs.
While I am waiting for test timing data (I need to find some more computers), have a blog post. blog.exe.dev/replace-your...
Replace your CI with a merge queue - exe.dev blog
Agent development means after-the-fact CI no longer works.
blog.exe.dev
Went hunting for why an end-to-end test was so slow. Found a wait loop that looked like it could be prodded by the test to speed it up. Doing so broke the test! Hours later, I have it: a subtle KVM entry/exit accounting bug in the production code. End-to-end tests are the way.
Went hunting for why an end-to-end test was so slow. Found a wait loop that looked like it could be prodded by the test to speed it up. Doing so broke the test! Hours later, I have it: a subtle KVM entry/exit accounting bug in the production code. End-to-end tests are the way.