SkillProof tested Claude skills

Scrape Analyze Page

Extracts every field + value from a saved HTML page, optionally against a JSON schema.

Door zytedata · zytedata/claude-skills

Werkt met setup ★ 8.0/10

Scrape Analyze Page — Extracts every field + value from a saved HTML page, optionally against a JSON schema.

Wat het doet

Cleans a locally saved HTML file, pulls structured metadata (JSON-LD/OpenGraph/microdata) via the extruct library, and merges it with visible page fields into a typed JSON output. Triggers only on already-downloaded files (rejects URLs) and supports a schema-guided strict extraction mode.

Testrapport

Ran it live on a fixture product page with JSON-LD: clean_html.py + extract_metadata.py (needs uv) correctly split visible DOM from structured metadata and the merged extraction matched a hand-done baseline field for field — but SKILL.md hard-codes `${CLAUDE_SKILL_DIR}/../scrape/references/python-environments.md`, a sibling-skill file that does not exist after a verbatim single-skill copy.

Getest op: 2026-07-15 · Claude Code 2.x (agent harness)

Installatie

git clone https://github.com/zytedata/claude-skills
cd claude-skills
mkdir -p ~/.claude/skills
cp -r skills/scrape-analyze-page ~/.claude/skills/scrape-analyze-page

Commando's en voorbeeldprompts

  • /scrape-analyze-pageExtracts every field + value from a saved HTML page, optionally against a JSON schema.

Skills reageren op gewone verzoeken — geen commando's om te onthouden. Na installatie activeren prompts zoals deze de skill (in het Engels):

  • Extract every field from this saved page product1.html into JSON
  • I downloaded this listing as listing.html, pull out every field and value
  • Analyze data.html against my schema.json in strict extraction mode

Volledig rapport (Engels) →