eeat-signals

name

eeat-signals

description

Audit and strengthen visible E-E-A-T signals and AI-search (GEO) readiness on Astro lead-gen sites. Use when a page or site needs Experience/Expertise/Authoritativeness/Trust signals, author credibility, first-hand-experience proof, trust pages (about/contact/privacy/returns), NAP + company registration display, review/trust-mark surfacing, AI-crawler access in robots.txt, or answer-first content for ChatGPT/Gemini/Perplexity/AI Overviews. Also triggers on "EEAT", "E-E-A-T", "GEO", "AEO", "AI search visibility", "get cited by AI", "trust signals", "author bio", "local trust", or making a small/niche business outrank larger competitors. Runs a deterministic auditor with runnable tests. Delegates JSON-LD generation to schema-entity-graph and JSON-LD validation to schema-audit; delegates copy voice to humanise-copy.

E-E-A-T Signals — audit + strengthen (Astro)

Make a page demonstrate Experience, Expertise, Authoritativeness and Trust to both humans and AI engines — and verify it with a runnable auditor. Trust is the load-bearing pillar; first-hand Experience is the asymmetric edge a real operator has over a larger competitor.

Scope boundary (read before doing anything)

This skill owns the human-visible + crawler-visible signal layer. It does not:

• generate JSON-LD → that is schema-entity-graph (schema-skill);
• validate JSON-LD / @id wiring / rich-result eligibility → that is schema-audit;
• rewrite copy tone/voice → that is humanise-copy;
• do build/perf/a11y/Lighthouse → that is astro-audit (plug EEAT in as a compliance check).

See BOUNDARIES.md for the exact seams. When a signal also has a schema form (author, reviews, sameAs, knowsAbout), check the visible side here and leave the markup to the schema skills. Never duplicate their rules.

The auditor (runnable)

npm install
npm test                 # 41 unit + integration tests
npm run audit -- --dir ./dist --market hu --robots ./dist/robots.txt
npm run audit -- --html ./dist/index.html --market uk

--market = uk | hu | generic. Exit code is 1 if any check is fail, else 0, so it can gate a deploy. The auditor only reports — it never edits the site. Statuses: pass / warn / fail / not_found (not-applicable, non-blocking). Gaps are surfaced as observed_differences, never as "opportunities".

DO — signals to place (then verify)

• Author: visible named byline on articles + a bio with specific credentials (years in trade, dates, certifications, real numbers) — not adjectives. (author.*)
• Experience: first-hand proof — case study, before/after, captioned own photos, first-person job description, original data. The one thing a competitor can't fake.
• Trust pages: about, contact, privacy always; terms; returns/refund/shipping for product pages. (trust-page.*)
• Business trust: visible phone, postal address, and company registration (UK Companies House number / HU cégjegyzékszám + adószám), plus recognised trust-marks. (business.*)
• GEO / AI: answer-first opening (~40–360 chars) + numeric facts; allow AI retrieval crawlers in robots.txt. (geo.*, ai-crawler.*)

Full detail: references/do-signals.md, references/ai-visibility.md.

DON'T — forbidden (these fail the audit or the review)

• Fake authors, AI-generated headshots, invented credentials, exaggerated experience.
• Scaled/mass-generated content; paraphrase-only pages with no added first-hand value.
• Fake/incentivised reviews; self-serving review markup (defer detail to schema-audit).
• Fabricated stats/case studies; claims of expertise with nothing demonstrated.
• Blocking AI retrieval bots while expecting AI visibility.
• Speakable schema on a non-news site; schema that doesn't match visible content.

Full detail: references/dont-forbidden.md.

Small business beats big competitor

Local + niche levels the field: genuine first-hand Experience, review velocity/recency, narrow topical depth, and proprietary data beat a national brand's weak local execution. Worked trapézlemez example: references/local-and-small-beats-big.md.

Market packs

• UK: Companies House number, Gas Safe / NICEIC / TrustMark / Checkatrade / FENSA / MCS / HETAS, UK directories. references/market-uk.md.
• HU: cégjegyzékszám + adószám (validated, incl. adószám CDV checksum), Árukereső Megbízható Bolt, Cylex / Arany Oldalak; Hungarian-language AI is under-served = first-mover edge. references/market-hu.md.

Conflict precedence

EEAT is an SEO-tier concern. If a signal conflicts with Safety/Legal, Accessibility, or Performance, those win (see the skills CONFLICT-MATRIX). Example: don't ship heavy author-photo galleries that blow the LCP budget — optimise first.

Workflow

1. Run npm run audit against the built site (or a page) with the right --market.
2. Fix every fail; weigh each warn; ignore not_found.
3. For any schema-expressed signal, hand off to schema-entity-graph / schema-audit.
4. Re-run until exit 0. Then run astro-audit for the full pre-deploy gate.

References

• BOUNDARIES.md — exact seams vs other skills
• references/do-signals.md
• references/dont-forbidden.md
• references/ai-visibility.md
• references/local-and-small-beats-big.md
• references/market-uk.md · references/market-hu.md