Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when auditing a page for E-E-A-T signals. The agent reads the page and scores Experience, Expertise, Authoritativeness, and Trustworthiness — then tells you exactly what to add to each dimension.
.claude/skills/inhouseseo-eeat-audit/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | -4% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 34% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 105% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 80% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 128% | 0% |
Scores a page on Experience, Expertise, Authoritativeness, and Trustworthiness — the four signals Google's quality raters use to evaluate content. Tells you what's missing and how to add it.
Real E-E-A-T is demonstrated, not declared. An author bio is table stakes. What matters is whether the content feels like it was written by someone who has actually done the thing.
URL of the page to audit. If the fetch fails, ask the user to paste the content directly.
You are a senior content quality evaluator with 10+ years reading for Google's quality rater framework. You can tell within 30 seconds of reading whether an author has done the thing they're writing about.
Fetch and read the full rendered page. Note everything that could be an E-E-A-T signal:
What you're looking for: evidence the author has DONE the thing, not just researched it.
Strong signals (8-10):
Weak signals (4-6):
Absent (1-3):
What you're looking for: accurate facts and depth beyond surface level.
Strong signals:
Weak signals:
Absent:
What you're looking for: does this content and author belong in the conversation?
Strong signals:
Weak signals:
Absent:
What you're looking for: transparency and honesty.
Strong signals:
Weak signals:
Absent (1-3):
| Signal | Score | Key Gap | |---|---|---| | Experience | /10 | | | Expertise | /10 | | | Authoritativeness | /10 | | | Trustworthiness | /10 | | | Total | /40 | |
Specific observations. "The screenshot in Section 3 is clearly from the author's own dashboard — this is a strong Experience signal."
Specific gaps with specific fixes:
Three changes you could make in under 30 minutes that would lift the E-E-A-T score materially. Ordered by impact.
Things that require more work but would fundamentally strengthen E-E-A-T: adding a methodology section, linking to related topical cluster pages, adding Author schema markup, creating an About page for the author.
To apply the fixes: use the improve-content skill with this URL, and paste the gap list as context.
Load from references/ only when the step calls for them.
Scoring and diagnosis:
ymyl-scoring-rubric.md — stricter scoring rubric for Your Money Your Life pages (finance, medical, legal) where the E-E-A-T bar is materially higher (Step 2, any YMYL page)experience-detection-playbook.md — how to tell in 30 seconds whether an author has done the thing (Experience dimension, when the page looks ambiguous)fastest-eeat-wins.md — ranked list of the highest-impact E-E-A-T fixes by implementation effort (Step 3, "Fastest Wins" block)eeat-signal-embedding.md — how to surface experience without a bio section or fake credentials (Step 3, "Structural Recommendations")author-schema-templates.md — copy-paste Person / Author / Organization JSON-LD for the schema fix (Step 3)YMYL content-type templates (references/content-types/) — load when auditing one of these types for the type-specific E-E-A-T bar:
thought-leadership.md, product-reviews.md, pricing-pages.md, service-pages.md, case-studies.md, about-pages.md| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 16,096 | 5,578 | -65% | 1 | 1 | 0% | 2,431 | 2,482 | +2% | 0 | 0 | — |
case-02 | fail→fail | 25,323 | 9,557 | -62% | 1 | 1 | 0% | 4,343 | 3,201 | -26% | 0 | 0 | — |
case-03 | fail→fail | 22,388 | 6,135 | -73% | 1 | 1 | 0% | 3,303 | 2,738 | -17% | 0 | 0 | — |
case-04 | pass→pass | 6,149 | 11,455 | +86% | 1 | 1 | 0% | 867 | 3,497 | +303% | 0 | 0 | — |
case-05 | fail→pass | 14,708 | 5,137 | -65% | 1 | 1 | 0% | 2,503 | 2,412 | -4% | 0 | 0 | — |
case-06 | fail→pass | 15,150 | 12,103 | -20% | 1 | 1 | 0% | 2,685 | 3,595 | +34% | 0 | 0 | — |
case-07 | pass→pass | 12,094 | 12,659 | +5% | 1 | 1 | 0% | 1,819 | 3,475 | +91% | 0 | 0 | — |
case-08 | fail→pass | 12,095 | 15,024 | +24% | 1 | 1 | 0% | 1,954 | 4,014 | +105% | 0 | 0 | — |
case-09 | pass→pass | 12,628 | 8,016 | -37% | 1 | 1 | 0% | 1,897 | 2,915 | +54% | 0 | 0 | — |
case-10 | pass→pass | 12,599 | 10,823 | -14% | 1 | 1 | 0% | 1,824 | 3,166 | +74% | 0 | 0 | — |
case-11 | fail→fail | 12,134 | 10,862 | -10% | 1 | 1 | 0% | 1,814 | 3,487 | +92% | 0 | 0 | — |
case-12 | pass→pass | 13,305 | 10,877 | -18% | 1 | 1 | 0% | 1,946 | 3,333 | +71% | 0 | 0 | — |
case-13 | pass→pass | 13,370 | 11,960 | -11% | 1 | 1 | 0% | 2,028 | 3,318 | +64% | 0 | 0 | — |
case-14 | fail→pass | 12,195 | 10,983 | -10% | 1 | 1 | 0% | 1,916 | 3,451 | +80% | 0 | 0 | — |
case-15 | fail→fail | 14,402 | 3,512 | -76% | 1 | 1 | 0% | 2,123 | 2,220 | +5% | 0 | 0 | — |
case-16 | pass→pass | 7,738 | 7,721 | -0% | 1 | 1 | 0% | 1,179 | 2,805 | +138% | 0 | 0 | — |
case-17 | fail→pass | 9,817 | 10,641 | +8% | 1 | 1 | 0% | 1,465 | 3,335 | +128% | 0 | 0 | — |
case-18 | fail→pass | 12,761 | 7,494 | -41% | 1 | 1 | 0% | 1,854 | 2,887 | +56% | 0 | 0 | — |
case-19 | pass→pass | 14,066 | 12,149 | -14% | 1 | 1 | 0% | 2,093 | 3,454 | +65% | 0 | 0 | — |
case-20 | pass→pass | 14,118 | 15,377 | +9% | 1 | 1 | 0% | 2,033 | 3,893 | +91% | 0 | 0 | — |
case-21 | fail→pass | 8,192 | 19,184 | +134% | 1 | 1 | 0% | 1,238 | 4,823 | +290% | 0 | 0 | — |
case-22 | pass→pass | 12,364 | 12,402 | +0% | 1 | 1 | 0% | 1,910 | 3,678 | +93% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +32 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.