Part 0 and Part A of the home-page redesign. Nothing under website/ is touched.
skills/
- design-taste-frontend installed to ~/.claude/skills/ as a VERBATIM copy of
github.com/Leonxlnx/taste-skill @ 717446e07a (2026-10-09, MIT). Body verified
byte-identical line by line; the only addition is a provenance block. The repo
holds scripts (skill.sh, scripts/*.mjs) but the SKILL FOLDER holds only SKILL.md,
so no script was copied and none was run. Not installed via npx or skill.sh.
- skills/felhom-web-design/SKILL.md is the override layer and names every conflict:
the static-HTML stack fence (no React/Tailwind/Motion/npm), nothing fetched from
another server (kills taste-skill's picsum/Unsplash/SimpleIcons ladder and its
npm icon libraries, since we have a self-hosted sprite), real images only, the
Hungarian gondolatjel surviving the en-dash ban, the claim rule, and the SEO-keep
list. 137 lines, check_skills.py PASS.
- SOURCES.md records the provenance and the kept/overridden table.
documentation/audits/redesign-2026-10-10/AUDIT.md
Measured against the LIVE site, not the file. 12 patterns to retire (R1-R12), each
naming the taste-skill rule it breaks; 7 to keep; dial reading 3/2/4 -> 6/3/4;
8 levers in priority order for the operator to approve at STOP 1.
Headline measurements: 10 922 px tall at 1440 and 19 913 px at 390 (23.6 phone
screens); 11 sections but only 8 layout families, why-grid used 3x and
apps-showcase 2x; 18 em dashes; 16 requests / 533 KB from exactly two hosts.
Contrast measures 7.56:1 body and 16.67:1 headings - AAA, and an accessibility win
to preserve rather than a thing to fix.
Method note: resize_window does not move window.innerWidth on this workstation and
DevTools device mode could not be driven, so the page was rendered in a same-origin
iframe at each width. Faithful, with two positive observables: the hamburger
computes to display:block at 390, and clientWidth == scrollWidth == 375.
Gates: site, hostinstall, hub-confirm, register and the rest green. instructions and
script-tests fail on this Windows workstation for environment reasons that predate
this change (E:\git\CLAUDE.md is the deliberate Windows adaptation the gate wants
identical; fcntl missing; relpath across C:/E:; the cp1250 console trap). Re-run on
DooPlex follows.
The four existing skills cover the product; nothing covered how work is
reported. Two rules this project has paid for — check the artifact rather
than the report, and do not state a claim more firmly than the evidence
allows — lived only in the operator's head and in chat, where Claude Code
never read them.
- felhom-evidence five confidence tiers, artifact-over-report
- felhom-diagnosis no hypothesis until a command has been seen red
- felhom-plain-language ASD-STE100, two options, the re-pitch
- felhom-handoff the note goes to a FILE, not the conversation
- felhom-doc-authoring the pointer decides whether material is reached
scripts/check_skills.py asserts what decides whether a skill is EVER
reached: frontmatter parses, name == directory, description and body
non-empty, under 150 lines, installed copy still samefile()s into the
repo. install_skills.py globs and never reads the file, so a missing
description installs perfectly and then silently never loads.
It convicted on its first run: felhom-build-deploy is 179 lines. NOT
trimmed here (pre-existing skills are out of scope, and trimming a
deploy skill without exercising its commands is how a wrong command
reaches a live host) — a named single-entry GRANDFATHERED exception,
WARNed every run, R-394. A new skill over the limit is convicted.
Red-proof run and seen failing: description removed from
felhom-evidence -> exit 1, "frontmatter field 'description' is missing
or empty". Restored, tree clean.
skills/SOURCES.md records both MIT upstreams, that these are adaptations
not copies, and the six pieces deliberately EXCLUDED with reasons.
Register: R-392 (no architecture doc covers the two-AI workflow),
R-393 (decision-log skill deferred, with the reason), R-394.
admin