Executive Producer of Slop. Journey before destination, tests before production.
Neverland
Joined February 2015
- Tweets2.2K
- Following655
- Followers1.3K
- Likes16.1K
This made me look back at @orta's Pragmatic Testing book that was lingering in my mind for quite some time. I turned the book into an agent skill and ran it on my iOS app. 378 tests went. Write up and skill in the thread 👇
🤖 Made with AI
OpenClaw deleted around 400k LOC of its own tests without much change in code coverage. Modern models just love writing tests for every tiny change, even if they aren't useful. This skill helped. github.com/openclaw/openclaw…
Same direction here: let code in, let the system grow, refactor often when you see things going out of hand. Only on personal projects so far tho. Commercially, I think it only works with a team that has extreme ownership and sense of responsibility for quality software.
I’ve turned off ai code reviewers. They don’t make sense anymore. Design and development and deployed state are all converging into the same thing.
Your implementation agent should be implementing correct code - external reviewers are redundant and add nuanced friction. Same reasons I’m bearish on external design software like figma/Paper. It costs the same to prototype in HTML as it does in your own code (probably cheaper given it’s real and non-redundant). Predict that paper will pivot to deployed-state canvasing.
More powerful/worthy of your token spend is external verifiers/automated-QA.
This take may change next week.
Sunday vibecoding project: I got tired of telling my agent "in Foo.swift, around line 40…"
so now in Xcode: select code → shortcut → type → Return
file, lines, code + my message go straight to the clipboard for Claude Code / Codex
Link in the tread 👇
@plannotator is still great for reviewing other people's code or annotating anything else. But when I'm reviewing code my agents wrote, I want to be in the IDE: the diff highlighting is better and it's much easier to navigate the surrounding code.
Serg Dort retweeted
原來 SF Symbols 27 有個 CLI
可以 search 跟 export
請給你的 agent 服用～
'/Applications/SF Symbols.app/Contents/Executa…'
My AI agent built the app with @xcodebuildmcp. Then Xcode had to rebuilt the same changes again. Configuring them to share build artefacts stopped the duplicate work. Here's the setup.
sergdort.github.io/xcode-age…
Mutation testing is GOATed
Did something similar yesterday for my iOS side project. My agent loved it
sergdort.github.io/refactor-…
Did something similar yesterday for my iOS side project. My agent loved it
sergdort.github.io/refactor-…
Very much looking forward
If your code sucks, your agent will too.
So, software fundamentals matter more than ever.
That's why I'm interviewing @unclebobmartin this Wednesday, live on YouTube.
Don't miss it. Link below.
Works great with @plannotator /plannotator-last
