seo-page analyzes a single supplied URL. It starts with a deterministic script, page_checks.py, that runs 47 on-page checks for titles, headings, canonical and robots signals, document basics, images, links, URL shape, page size, social tags and JSON-LD, and returns findings with stable ids and evidence.
The agent then decides which findings matter for that page, adds content quality, E-E-A-T, schema and image analysis, and flags likely LCP, INP and CLS problems. The result is a score card, issues sorted by priority and ready-to-use JSON-LD.
What You Get
- A score card out of 100 with five sub-scores
- Issues grouped Critical, High, Medium and Low, each with quoted evidence
- Specific recommendations with expected impact
- JSON-LD snippets for the schema opportunities it found
Common Use Cases
- Check a landing page before it goes live
- Explain why one article is not ranking
- Review a page saved as HTML when it is behind a login
- Spot-check titles, canonicals and noindex on a client page
What Makes It Unique
Counting is done in code and judgment is left to the model. Every length and count comes from the script’s JSON, checks that could not run are reported as not checked, and length limits are described as truncation heuristics because Google publishes none. It also uses the format a server delivers when it judges images.
Under The Hood
The skill runs the plugin’s Python scripts through a launcher: page_checks.py to analyze a URL or a saved HTML file, fetch_page.py to render a JavaScript page, parse_html.py to probe images over the network, and metadata_template.py to compare title and description. These fetch pages from the target site and may write the rendered page to a file. With DataForSEO connected it can also request SERP and backlink data. The frontmatter declares none of this access. The skill states that script output quotes the page and must be treated as untrusted data, as must all fetched content.