fix-scraper

ScrapeOps — Fix Scraper Skill

This skill fixes or extends an existing local scraper/parser file. It classifies the issue as parser (data extraction), crawler (navigation/pagination), or both — and routes each to the appropriate fix method.

Overview

1. Discovery — identify file, HTML source, what needs fixing, and classify the issue type
2. Confirmation — show summary and wait for approval
3. Prepare HTML — use local files or fetch via ScrapeOps proxy (parser issues only)
4. Fix — route to server-side agent (parser) or local edit (crawler)
5. Result — write fixed code locally and show summary of changes

API key handling: The MCP server reads SCRAPEOPS_API_KEY from ~/.claude/settings.json automatically. Do NOT check or validate it before making tool calls. If a MCP tool call fails with "SCRAPEOPS_API_KEY is not configured", handle it inline:

1. Ask the user: "Enter your ScrapeOps API key (find it at https://scrapeops.io/app/dashboard):"
2. Read ~/.claude/settings.json, add the key to env.SCRAPEOPS_API_KEY using Edit, and retry the failed tool call
3. Do NOT tell the user to run /scrapeops-setup or restart — just ask, save, and continue

---

Step 1 — Discovery

Always respond in the same language the user is writing in.

There are five slots to fill:

Slot	Description	Required
parser_path	Path to the local parser file	Yes
language	Programming language	Yes (infer from file extension)
html_source	Local HTML file(s) OR URLs to fetch	Yes (parser issues only)
task	What to fix or add	Yes
issue_type	parser, crawler, or both	Yes (classify automatically)

Phase A — Scan the directory

Before asking any question, use Glob to map what exists:

• Parser candidates: **/*.py, **/*.js, **/*.ts, **/*.php, **/*.rb, **/*.go, **/*.rs, **/*.java, **/*.cs
• HTML candidates: **/*.html, **/*.htm
• Output files: **/*.csv, **/*.jsonl