@zeroskillz@bcherny@AnthropicAI Do u have disable_telemetry enabled? If you do it will not enable most of the features they are slowly rolling out like /remote-control or /adviser, I assume because they are a/b testing
@RyanCarniato TDD is just the start. Strict linting, complexity limits, file length caps, typechecking, duplicate detection all the "too much ceremony" stuff is now completely worth it because it helps the LLM reason about your code and saves tokens.
the explicit nature off tailwind and getting rid of inheritance is one of the major draws of tailwind.
you have to lean into the copy paste duplication because it's explicit. maybe it's worse to write initially but when you are tying to maintain or trouble shoot why something works you'll appreciate it.
Haven't all the flagship model providers pulled similar bait-and-switches? I remember when Claude 4 launched. It was incredible for the first week or two, then performance gradually degraded. They likely either scaled back compute resources or hit scaling bottlenecks.
It wouldn't be surprising if OpenAI is doing the same thing on a larger scale, giving influencers and partners advanced early access to generate hype, then throttling performance for general users.
Have spent time with claude code? Cursor feels like an unneeded abstraction once you do.
$20 Cursor is worth it for the tab completion, but the more I use cc the less I even use that honestly.
Pretty much just use the IDE to explore the changes its making then back to cc unless it a very minimal manual change
1. read everything you can that anthropic has published in their docs:
https://t.co/jd1q2DhCeY
2: start small. do a small green field project. use a planning prompt. do not turn on auto accept until you have held its hand for a while. say no often and correct what it's doing. once it's on the right track then let it go an auto accept or even --dangerously-skip-permissions
3: connect it to gh cli tool and have it branch, commit and push changes
4: playwright MCP is the only MCP you need. have it tdd or at least verify/validate it self via the cli
5: read https://t.co/QXd6o3HUBG again
6: understand claude code's philosophy. raw modal access
7: find and watch all the @bcherny interviews you can https://t.co/FGKk3yGtf8
AI's superpower is translation. In this case it lowered the cost of understanding making it worth my time.
I'll admit it wasn't that long ago I would have been really put off by some of the "mixing of religion and tech" coming from some of the smartest tech influencers I know.
But it has helped burst my bubble a little bit. Then when you look back in history and realize what organized religion has done for art and science and it's like of course...