Dev finds 5 performance disasters in Claude-generated code, fixed by vercel-react-best-practices skill
julianharris · x · 2026-11-06
Developer Julian Harris shares astonishingly bad Opus-generated sandbox code: on every token, the whole conversation (generated HTML included) is synchronously serialized to localStorage — O(n²) work that blocks the page and silently fails past the 5MB quota. Screenshot lists 5 issues total, including full-app re-renders on every keystroke/token and GET requests polluting history. He credits the /vercel-react-best-practices skill for the cleanup and asks which skills others rely on.
More from coding & agent
- GitHub rebuilds Git storage for thousands of concurrent AI agents — arpit_bhayani · 2026-11-06
- OpenAI built Pando for Codex: 5x lower Git push latency, ~1PB transferred daily — sven_ai · 2026-11-06
- Snorkel AI: agents peek at reward code ~25% of the time rather than asking users — ajratner · 2026-11-06
- Subagents fail because summaries are lossy — checkpoint agents fork context instead — narphorium · 2026-11-06
- Building a pure-code lightbulb with Claude: WebGL shader, MediaPipe tracking and self-made test videos — ramramlab · 2026-11-06
- PEARtopia lets your Claude, Gemini and Cursor agents play a MULE-style board game with you — ctenidae8 · 2026-11-06