| name | web-scraper-generic |
| description | Use when the user wants to extract content from any website not covered by a dedicated site skill — extract visible text, links, tables, structured metadata (JSON-LD, Open Graph), or take a screenshot-based reading of the current page. Works on any public URL without prebuilt action code. A real browser is required for JavaScript-rendered pages; plain fetch works for static HTML. |
Generic Scraper — Fallback Scraping Skill
Use this skill when the target website has no dedicated per-site skill. These actions extract whatever is on the page using general-purpose browser automation — no prebuilt selectors, no site-specific logic.
When to use
- Site not in supported list → use generic scraper
- Need quick text dump of a page →
extract-text
- Need all links/URLs from a page →
extract-links
- Need structured data (JSON-LD, meta tags) →
extract-structured
- Need tables from a data page →
extract-tables
Action Index
- extract-text — Extract all visible text content from the current page, preserving heading hierarchy. Best for article/blog/content pages. Full spec: references/extract-text.md.
- extract-links — Extract all hyperlinks from the page with their text labels and URLs. Full spec: references/extract-links.md.
- extract-structured — Extract JSON-LD, Open Graph, Twitter Card, and meta tag data from the page head. Best for pages with rich metadata. Full spec: references/extract-structured.md.
- extract-tables — Extract all HTML tables from the page as arrays of row objects. Full spec: references/extract-tables.md.
- extract-csv — Find CSV download links and inline tabular datasets; parse HTML data tables into structured row objects. Full spec: references/extract-csv.md.
