Redesign an existing website to make it better. Built on top of impeccable.
88%
Does it follow best practices?
Run evals on this skill
Adds up to 20 points to the overall score
Low
Low-risk findings worth noting
rollout:baseline sourceThese are the deterministic detectors of the rollout:baseline source that
optimize.mjs runs over the delivered (or migrated) HTML. Baseline is one of
several audit sources — the others (impeccable, the marketing SEO skills,
stardust tensions) feed the same ledger via findings.mjs record; see
audit-sources.md. Findings reference a check by layer + check; this catalog
can grow without a schema change. Severity drives the gate (any open P1 fails
it); fixability routes the fix; a registered check is auto-fixed by autofix-aem.
fixability: PM = platform-migration (rollout re-deploys to fix) · DP =
design-pass (upstream — fix in migrate/prototype; rollout only surfaces) · OOS
= out-of-scope (informational).
| check | sev | fix | fires when | recommended move |
|---|---|---|---|---|
landmark-main | P1 | PM | no <main> in the delivered HTML | ensure block output lands inside <main> (deploy anti-pattern 17) |
img-alt | P2 | DP | one or more <img> without alt= | author alt text upstream; rollout can't synthesize it |
| check | sev | fix | fires when | recommended move |
|---|---|---|---|---|
title-missing | P1 | PM | no <title> | add a metadata block; EDS derives <title> (deploy #34) |
title-length | P3 | PM | <title> < 10 or > 70 chars | ~50–60 chars: brand + primary keyword |
meta-description | P2 | PM | no <meta name="description"> | add a Description row to the metadata block |
single-h1 | P1 | PM | <h1> count ≠ 1 | hero headline = single <h1>; others <h2> (deploy #35) |
canonical | P2 | PM | no <link rel="canonical"> | emit a self-canonical link at delivery |
sitemap (site) | P2 | PM | rollout/site/sitemap.xml absent | run assemble.mjs and deploy the sitemap |
| check | sev | fix | fires when | recommended move |
|---|---|---|---|---|
jsonld | P2 | PM | no application/ld+json | emit page-type JSON-LD in the head |
| check | sev | fix | fires when | recommended move |
|---|---|---|---|---|
duplicate-title | P2 | DP | ≥2 pages share a <title> | unique titles upstream (migrate metadata) |
duplicate-description | P3 | DP | ≥2 pages share a description | per-page descriptions upstream |
brand-tensions, design-ux, and content-conversion need judgment, not
deterministic parsing, so the baseline source leaves them unscored (null).
They are populated by the other sources — impeccable:critique (design-ux,
brand-tensions), stardust:tensions (brand-tensions, design-ux), and the
marketing skills (content-conversion via cro/copywriting if run) — recorded
through findings.mjs. Once a source records into a layer, the scorecard scores
it; until then it shows not-assessed rather than faking a number.
When the page's capture exists (stardust/current/pages/<slug>.json plus its
rendered sidecar pages/<slug>.html; optimize.mjs --current <dir> points
elsewhere), a baseline finding whose condition the SOURCE shares is not a
delivery defect:
| check | tagged when |
|---|---|
title-missing | the source page has no <title> either |
title-length | the delivered <title> equals the captured one (entities decoded, whitespace collapsed) |
meta-description | the source page has no meta description either |
jsonld | the source's rendered DOM has no application/ld+json either (no sidecar → unknown → not tagged) |
duplicate-title / duplicate-description | the captures of the same pages share the delivered value |
The findings schema has no parity property, so the tag is fixability: out-of-scope (the schema's own "informational" value) plus the evidence prefix
source parity: <evidence> — <shared condition>; autofix is set unavailable (a
faithful migration is never rewritten). Tagged findings stay open under their
stable id, are listed in the report's own "source parity" section, and are
excluded from every dimension score, from the open P1/P2/P3 counts and from the
gate. The tag follows each run: a finding the source stops sharing (a later
capture, a changed title) routes normally again. A recorded hands-off run
accepted 63 of these by hand — every JSON-LD, title-length, meta-description and
duplicate-title finding of a faithful migration.
id-keyed by layer|check|page-slug; site-level by
layer|check|* (or the involved slug set). Re-runs match by id — no duplicates.fixed) when this run actually inspected its
scope (the page was loaded, or site checks ran). A --slug run never marks
another page's findings fixed.accepted / wontfix are human decisions and are never auto-reopened..tessl-plugin
skills
audit
reference
deploy
reference
scripts
test
diff
direct
dynamics
extract
migrate
prepare-migration
prototype
assets
motion
fixtures
reference
qa
replica
reference
scripts
test
reskin
rollout
spec
stardust
uplift
reference