# Scrape

> Read-only structured extraction from pages, tables, listings, or web apps using Browser, Chrome, web fetch, or repo-local evidence; return JSON or tables.

- Skill: `mphaxise/scrape` (Agent Skill)
- Install (CLI): `npx skillmds@latest add mphaxise/scrape`
- Raw SKILL.md: https://api.skillmd.com/api/skills/mphaxise/scrape/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: mphaxise (https://skillmd.com/u/mphaxise)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/mphaxise/scrape

---


# Scrape

Use this skill when the user asks to pull data from a page, table, listing, or web app.

## Workflow

1. Clarify the target URL and desired fields if not obvious.
2. Use the safest read-only inspection path:
   - `@Browser` for public/local unauthenticated pages
   - `@Chrome` only when signed-in browser state is required and approved
   - web fetch or repo-local fixtures when rendered interaction is unnecessary
3. Extract only the requested data.
4. Return JSON or a table with source notes.
5. If the flow is useful and repeated, suggest `skillify` to preserve it as a reusable browser skill.

## Guardrails

- Treat scraping as read-only.
- Respect authentication boundaries and site terms.
- Do not fabricate missing fields.
- Include selectors, URLs, or evidence notes when the extraction may need to be repeated.