Visual regression page for the current tweet embed system, including cache-backed renders, URL normalization, note tweets, media, and article cards.
Visual regression page for the current tweet embed system. These examples exercise the same cache-backed render path the public site now ships: scripts/sync-tweets.mjs fetches syndication JSON during prebuild, writes src/generated/tweet-cache.json, and the <Tweet /> component renders from that cache first instead of trusting prerender-time network luck.
Doing my daily viewport gymnastics 🤸♂️
"Instead of making design decisions on strict, limited breakpoints, keep in mind the sheer amount of fragmentation there is in viewports."
Good reading:
viewports.fyi
Doing my daily viewport gymnastics 🤸♂️
"Instead of making design decisions on strict, limited breakpoints, keep in mind the sheer amount of fragmentation there is in viewports."
Good reading:
viewports.fyi
Got flooded with DMs asking for the markdown file.
Here it is, bookmark this.
---
name: cost-estimate
description: Estimate development cost of a codebase based on lines of code and complexity
---
# Cost Estimate Command
You are a senior software engineering consultant
Fun command built in Claude Code: /cost-estimate It scans your codebase and cross-references current market rates to calculate what your project would've cost a real team to build. It looks at all the APIs, integrations, everything. Without AI: ~2.8 years. ~$650k. With AI:
Got flooded with DMs asking for the markdown file.
Here it is, bookmark this.
---
name: cost-estimate
description: Estimate development cost of a codebase based on lines of code and complexity
---
# Cost Estimate Command
You are a senior software engineering consultant
Fun command built in Claude Code: /cost-estimate It scans your codebase and cross-references current market rates to calculate what your project would've cost a real team to build. It looks at all the APIs, integrations, everything. Without AI: ~2.8 years. ~$650k. With AI:
Had breakfast with a founder this morning who builds software for independent moving companies.
He was in NJ to meet a customer, and DM'd me on X. I had never met or talked to him before, but the conversation was super interesting.
There are about 7,000 of them in the US. Most
GPT 5.4 is the only model where the more context (the right context) and steering you give, the better it performs on long horizon tasks. Where Anthropic models currently turn to absolute moosh except in *just right* circumstances Codex + 5.4 is EATING this stuff up.
Notice
