From 9a250406bc53f29544fe05f6bac820ec2f1b4887 Mon Sep 17 00:00:00 2001 From: Claude Date: Fri, 26 Jun 2026 13:23:52 -0300 Subject: [PATCH] docs: show the web-skill improvements on the landing page Expand the "score -> fix -> re-score" cards to a 2x2 of real published-skill turnarounds, led by daydream (no frontmatter, 38 -> 100, +62 Reject -> Ship): daydream +62, annotate +50, bad-skill +27, rehydrate +8. Lead now headlines "turnarounds up to +62, Reject -> Ship". Co-Authored-By: Claude Opus 4.8 (1M context) --- docs/index.html | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) diff --git a/docs/index.html b/docs/index.html index 897ddfa..8630224 100644 --- a/docs/index.html +++ b/docs/index.html @@ -188,12 +188,16 @@

Benchmark

docxfile_transform92 / A599 lines & ~5142 tokens (over budget)

The upgrade: score → fix → re-score

-

skval doesn't just grade — its ranked findings drive the fix. The biggest real-world turnarounds:

+

skval doesn't just grade — its ranked findings drive the fix. Real published skills, scored verbatim then fixed and re-scored to a clean 100 — turnarounds up to +62, Reject → Ship:

+

daydream · no frontmatter · 38100

+

A real published skill (glebis/claude-skills) shipped with no YAML frontmatter — no name, no description, so it can't trigger. Add the frontmatter block: compare.py +62, Reject → Ship. case study →

annotate · invalid frontmatter · 50100

-

A real published skill (glebis/claude-skills) whose unquoted colon broke its YAML — it used to crash skval, and now scores. One-line fix: quote the description. compare.py: +50, Revise → Ship. case study →

+

An unquoted colon broke its YAML — it used to crash skval, and now scores. One-line fix: quote the description. compare.py +50, Revise → Ship. case study →

bad-skill · 73100

-

Four findings fixed one-for-one — kebab name, drop <>, broken ref, stray key. compare.py: +27, Revise → Ship. case study →

+

Four findings fixed one-for-one — kebab name, drop <>, broken ref, stray key. compare.py +27, Revise → Ship. case study →

+

rehydrate · <> in description · 92100

+

Angle brackets in the description break parsers and triggering — drop them to score clean. compare.py +8. case study →