I scored this repo for AI-Readiness and it got 38%, #18008
Replies: 1 comment 1 reply
|
Great timing on this — the AI-readiness scoring concept captures something real. Structured project context (whether YAML, CLAUDE.md, or the GEMINI_SYSTEM_MD format) makes a measurable difference in output quality. A layer worth adding beyond project metadata: the behavioral rules that prevent version-specific drift. For Svelte 5 specifically, AI tools trained mostly on Svelte 4 examples will default to the old patterns without explicit guidance: These seven rules eliminate the most common Svelte 5 drift we have seen. The project metadata tells the AI what you are building; the behavioral rules tell it which patterns are allowed. We published a free Svelte 5 CLAUDE.md starter with the full 15-rule set: https://gist.github.com/oliviacraft/2d7f21b68741593b292c01401772ca01 |
Uh oh!
There was an error while loading. Please reload this page.
Inspired by @MathiasWP's story 👏, decided to share this here.
I scored this repo for AI-Readiness and it got 38%, so I built a tool to improve the score, and now it gets 100%. The two files are below — now AI knows what I'm working on and is optimized for my exact project.
Before (38%)
After (100%)
All reactions