Images can be discovered through Google Search, Google Images, and visual-search products, while image delivery can materially affect page performance. Claude SEO v1.8.1 brought image-result analysis and file optimization into the same terminal workflow.
Claude SEO v1.8.1 changes this. It introduces Google Images SERP analysis that shows exactly which domains dominate image results for your keywords, automated image file optimization that converts to WebP/AVIF and injects IPTC metadata, and a cross-skill that connects your on-page images to live SERP data. The result: you can identify image ranking opportunities, optimize files for those opportunities, and track your image presence, all from the terminal.
Run /seo dataforseo serp-images "best seo tools" to see which domains own the image results, what formats they use, and where your opportunity gaps are.
What does Google Images SERP analysis reveal?
The new /seo dataforseo serp-images command fetches live Google Images results for any keyword via the DataForSEO API. Instead of guessing which images rank, you get concrete data across four dimensions.
Domain presence analysis shows which websites appear most frequently in the captured image results. The distribution depends on the query, locale, device, and collection date, so the report presents observed counts instead of assuming a universal concentration benchmark.
Alt text pattern analysis reviews the attributes found on pages represented in the captured results. These observations can inform accessible descriptions, but they do not reveal a ranking formula or justify copying another page's phrasing.
Format distribution breaks down whether observed images use WebP, JPEG, PNG, AVIF, or another supported format. Modern formats can reduce transferred bytes, but format alone is not an image-search ranking advantage.
Opportunity review identifies queries where the site appears in web results but its images are absent from the captured image results. Treat these as investigation candidates, then inspect the page, image relevance, crawlability, and current result set.
/seo dataforseo serp-images "technical seo audit"
How does image file optimization work?
The /seo images optimize command handles the full image optimization pipeline locally, without any API keys. It runs four stages automatically:
- Format conversion - Converts JPEG and PNG files to WebP (25-34% smaller than JPEG at equivalent quality, per Google's WebP compression study) or AVIF (over 50% smaller in some cases, per web.dev). Original files are preserved as backups
- IPTC/XMP metadata injection - Embeds Creator and Copyright fields into image files. Google reads these fields and can display them in Google Images rich results
- Responsive variant generation - Creates multiple sizes (480w, 768w, 1024w, 1440w) for srcset attributes, so browsers load the right size for each device
- Metadata audit - Reports what metadata already exists, what was injected, and confirms the output files are valid
/seo images optimize ./assets/blog/
Images are frequently the Largest Contentful Paint element, according to the HTTP Archive Web Almanac 2025. Responsive dimensions, suitable compression, and correct loading priority can improve performance when an image is the bottleneck. Google's image SEO guidance recommends accessible, descriptive context and supported formats, but it does not publish a weighting formula.
What is IPTC metadata and why does it matter?
IPTC (International Press Telecommunications Council) metadata is a standard for embedding structured information directly in image files. Unlike alt text, which lives in HTML, IPTC data travels with the image file itself. When Google crawls your images, it reads these embedded fields alongside your page content.
Google reads 5 specific IPTC fields, according to IPTC.org's guide for Google Images:
- Creator - The photographer or creator name. Google displays this in image search results
- Copyright Notice - The copyright holder. Establishes content ownership when images get hotlinked
- Credit Line - How the image should be credited
- Web Statement of Rights - Required for Google's "Licensable" badge in image results (launched August 2020)
- Digital Source Type - New field for identifying AI-generated images, increasingly relevant for content authenticity
Google's image license metadata documentation confirms that IPTC metadata is read during indexing. Images with proper Creator and Web Statement of Rights fields can appear with a "Licensable" badge in Google Images, giving them more visual prominence than plain results.
Claude SEO injects these fields automatically using ExifTool during the optimization pipeline. You provide the creator name once, and it applies to all processed images.
Required dependency
The image optimization commands require ExifTool and the WebP tools package:
sudo apt install libimage-exiftool-perl webp
How does the image SERP cross-skill connect everything?
The /seo images serp command bridges your on-page image audit with live SERP data. It cross-references the images on your pages against Google Images rankings to answer three questions:
- Which of your images already rank? - Identifies images that appear in Google Images results, so you know what's working
- Where are the gaps? - Finds keywords where you rank organically in web search but have no image presence
- What formats win? - Shows which image formats perform best for your target queries, so you can prioritize conversion
/seo images serp "seo audit tool"
This connects to the existing image optimization skill that already checks alt text, file sizes, lazy loading, and CLS prevention. The SERP cross-skill adds the competitive layer that was previously missing.
What else changed in v1.8.1?
Beyond the image features, v1.8.1 includes important bug fixes and standards compliance work. These changes improve portability and maintenance:
| Fix | Details |
|---|---|
| Version mismatch | Unified all 19 SKILL.md files, plugin.json, and CLAUDE.md. Was a three-way split between 1.7.0, 1.7.2, and 1.8.0 |
| Broken reference path | seo-backlinks now correctly points to shared backlink-quality.md reference file |
| Hardcoded paths | Removed absolute ~/.claude/skills/ paths from agents and skills. Now plugin-relative for portability |
| Skill file size | Moved seo-dataforseo tool catalog to a reference file, reducing from 416 to 380 lines |
| Table of Contents | Added TOC to prompt-engineering.md (326 lines, required by Anthropic standard at 300+ lines) |
The entire codebase was audited against the Anthropic Skill Creator Standards. All 19 skills pass YAML frontmatter validation, stay under 500 lines, follow progressive disclosure, and include proper error handling. Overall compliance score: 88/100.
How does this fit into a complete image SEO workflow?
Here is a practical workflow using the new commands alongside the existing site audit skill:
- Run
/seo imagesto audit all images on your site for alt text, file sizes, formats, and lazy loading - Run
/seo images optimize ./images/to convert to WebP, inject IPTC metadata, and generate responsive variants - Run
/seo dataforseo serp-images "your keyword"to inspect the domains and formats observed in a dated image-result snapshot - Run
/seo images serp "your keyword"to cross-reference your images with live SERP data - Use the opportunity scores to prioritize which pages need image content improvements
The technical SEO skill also reviews image delivery. Responsive variants can reduce unnecessary bytes, and explicit dimensions can prevent layout shifts. Actual Core Web Vitals impact must be verified with field data when coverage is available.
Install or upgrade
New install:
claude /install github:AgriciDaniel/claude-seo
Install optional dependencies for image optimization:
sudo apt install libimage-exiftool-perl webp
The image optimization commands work without any API keys. Only the Google Images SERP analysis requires the optional DataForSEO extension, which was introduced in v1.7.2.