Idea
Generate a compact llms.txt (or .cursor/rules-style file) distilled from docs/llm-guide.md, for end users to drop into their own AI coding assistant's context (Claude Code, Cursor, Windsurf, etc.) when building sites with SLASHED. This gives their AI tool accurate, current-version knowledge of the token/class vocabulary instead of guessing or hallucinating class names.
Why
docs/llm-guide.md is already a maintained, CI-gated reference (npm run check:llm-guide) with per-domain token sections and naming conventions — most of the content work is done. This is packaging/distillation, not new research: a build script that extracts the token/class vocabulary into a shorter, assistant-optimized format and ships it as a downloadable/distributable artifact (e.g. alongside badges/ in the release, or via a doc page).
Notes
- Should be regenerated by the existing
npm run docs pipeline so it can't drift from llm-guide.md.
- Consider adding to the
release.yml artifact set alongside dist bundles.
Idea
Generate a compact
llms.txt(or.cursor/rules-style file) distilled fromdocs/llm-guide.md, for end users to drop into their own AI coding assistant's context (Claude Code, Cursor, Windsurf, etc.) when building sites with SLASHED. This gives their AI tool accurate, current-version knowledge of the token/class vocabulary instead of guessing or hallucinating class names.Why
docs/llm-guide.mdis already a maintained, CI-gated reference (npm run check:llm-guide) with per-domain token sections and naming conventions — most of the content work is done. This is packaging/distillation, not new research: a build script that extracts the token/class vocabulary into a shorter, assistant-optimized format and ships it as a downloadable/distributable artifact (e.g. alongsidebadges/in the release, or via a doc page).Notes
npm run docspipeline so it can't drift fromllm-guide.md.release.ymlartifact set alongside dist bundles.