/seo-content

name: seo-content
description: Content quality and E-E-A-T analysis with AI-citation-readiness assessment. A deterministic eight-dimension audit (author / credentials / dates / sourcing / first-hand signals with a stricter YMYL bar, structure, readability, depth by page type, keyword placement, links, scaled-content tells like leaked template placeholders, passage citability) scores a page 0-100 with a fix per finding, then the skill judges what a script can't. Trigger when the user says "content quality", "E-E-A-T", "content analysis", "readability check", "thin content", "content audit", or "is this content good enough".

seo-content

**Family:** seo **Status:** Stable

Purpose

Evaluate a piece of content for both human trust (E-E-A-T) and machine citability. The dual framing matters in 2026: content must convince human reviewers *and* be the kind of passage an LLM will quote.

Triggers

• "content quality" / "content audit"
• "E-E-A-T" / "readability" / "thin content"
• "is this content good" / "AI citation readiness"

Inputs

• Content URL or local markdown/text/HTML file
• Target audience and search intent
• Author / publication signals (if relevant)

Steps

1. **Run the content audit** (HTML, Markdown or text; `--type` sets the depth floor; YMYL is auto-detected — force with `--ymyl yes|no`; `--as-of` enables staleness):

   python3 "${CLAUDE_PLUGIN_ROOT}/scripts/seo/content_audit.py" --file <file> --url <page URL> \
     --keyword "<primary topic>" --type article --as-of <YYYY-MM-DD> --human

Eight dimensions, every finding with a fix, plus a 0-100 content score — rubric, thresholds and scoring in `references/seo-content/content-rubric.md`:

• **E-E-A-T signals** — author (meta / schema / byline), credentials or expert

review (required on YMYL), publish + updated dates and staleness, statistics without an outbound source, first-hand experience markers, About/Contact.

• **Structure** — one H1, no heading-level skips, H2 sections on long content.
• **Readability** — sentence length, very long sentences, walls of text, Flesch.
• **Depth** — word count against the page type's floor; lists/tables on long pages.
• **Keyword** — title / H1 / intro / subheading / meta / slug placement, stuffing.
• **Links** — in-content internal links (nav/footer excluded), generic anchors.
• **Originality** — leaked template placeholders (critical: a scaled-content tell),

generic filler phrasing, duplicated sentences. Never labeled "AI-written".

• **Citability** — passage citability via `geo_check.py` (same scorer as seo-geo).

2. **Judge what a script can't:** is the first-hand experience real, are the credentials relevant to the topic, is each claim *correct*, does the page answer the intent better than what already ranks? Read against `references/shared/eeat-criteria.md`. 3. **Fix the weak passages** listed under `citability` (the full list comes from `geo_check.py --content <file> --human`) so each leads with one specific, sourced claim. 4. **Render** the score, per-dimension findings grouped Critical → Info, and sentence-level revisions — label observable signals as such, never "Google's E-E-A-T score".

Outputs

• Content score (0-100) with per-dimension counts and every finding's fix
• YMYL verdict and the raised-bar findings it triggers
• Sentence-level revision suggestions, including the weak-passage list

Dependencies

• `scripts/seo/content_audit.py` (required) — the eight-dimension content audit
• `scripts/seo/geo_check.py` (required, imported) — passage citability scoring
• `references/seo-content/content-rubric.md` (required) — thresholds, YMYL bar, scoring
• `references/shared/eeat-criteria.md` (required) — the E-E-A-T concept the signals serve
• Related: `seo-geo` (AI-search depth), `seo-content-brief` (production counterpart),

`content-draft` (writes the draft) — composed, not hard-depended (one-directional graph)

Notes

This skill audits; `content-draft` produces and `seo-content-brief` plans. Citability is now a first-class quality dimension, not a nice-to-have.