<< All versions

Skill v1.0.0

currentLLM-judged scan90/100
oegeyilmaz9/seo-aeo-geo-ultimate/seo-images
──Details
PublishedSeptember 27, 2026 at 11:43 AM
Content Hashsha256:60d78e2575002194...
Git SHA
──Files
Files (1 file, 3.9 KB)
SKILL.md3.9 KBactive
SKILL.md · 43 lines · 3.9 KB

version: "1.0.0" name: seo-images description: Use when reviewing image accessibility, discoverability, metadata, responsive delivery, image sitemaps, or visual performance from a page/media inventory; provide evidence-linked improvements without keyword, format, or ranking guarantees.


SEO Images

Purpose

Improve the way meaningful visual assets are understood and delivered to people and crawlers. This skill separates accessibility, rendering/performance, asset rights, and image-search discoverability instead of treating them as one score.

Read references/image-review-protocol.md before preparing a formal media recommendation.

Input gate

Require the rendered page or asset inventory, image purpose/context, source URL, intrinsic dimensions/format/size, delivery behavior, locale, accessibility context, and any available field/lab evidence. Ask who owns the image/license and whether an image is content, UI decoration, hero/LCP candidate, product/media asset, or purely decorative.

Do not infer file format, dimensions, loading behavior, or user impact from a filename. Do not recommend an image or alt text without seeing its page context.

Workflow

  1. Inventory by purpose. Map each material image to the visible topic, page region, adjacent text/caption, owner/license, and delivery URL. Mark decorative images as decorative rather than inventing alt text.
  2. Review accessibility. Check meaningful alternative text, captions/nearby explanation, controls, contrast/legibility where text is embedded, and responsive mobile rendering. Alt text explains the image’s purpose in context; it is not a keyword field.
  3. Review delivery. Inspect dimensions, responsive candidates, intrinsic width/height or aspect-ratio reservation, loading priority, lazy loading behavior, cache/CDN policy, and actual LCP/CLS/interaction evidence when available. A newer format is not automatically an improvement; evaluate support, source quality, byte savings, and visual fidelity.
  4. Review discovery/metadata. Check whether significant images are reachable in rendered content and whether page context, canonical URLs, structured data, and image sitemap use are truthful and relevant. Route structured data to seo-schema and sitemap changes to seo-sitemap.
  5. Plan safe changes. For each action name the asset/template scope, owner, accessibility/content review, visual QA, performance test, approval, and rollback. Send systemic delivery changes through seo-action-plan.

Guardrails

  • Do not use keyword-stuffed alt text, duplicate captions, decorative image descriptions, fake EXIF metadata, or image sitemap stuffing.
  • Do not promise image search visibility, page ranking, Core Web Vitals, or conversion improvement.
  • Respect licenses, consent, privacy, and brand requirements. Do not expose internal image URLs or generate derivative assets without authorization.
  • Keep image discovery, user accessibility, and rendering performance as separate verifications.

Formal evidence handoff

When this work needs a cross-team, approval-ready plan, package evidence-bound findings as an immutable seo-findings.json bundle using the checked-out suite contract. Let <suite-root> mean ${CLAUDE_PLUGIN_ROOT} in Claude Code. In Codex, read .seo-suite-runtime.json beside this SKILL.md when present and use its suite_root value; otherwise use the absolute repository checkout. Keep every referenced capture/source below raw/, retain declined claims and limitations, and run python "<suite-root>/scripts/validate_seo_findings.py" validate-findings <bundle>/seo-findings.json --bundle <bundle>. Send only a passing bundle to seo-action-plan; otherwise label the handoff provisional.

Output

Return a media inventory with observations, accessibility/performance/discovery recommendations, source-aware implementation notes, test path, and rollback. Provide alt-text or caption options only with visible context and label them as reviewable suggestions.

All versions