jevcrawl vs Scrapewise

jevcrawl

Three compact, typed MCP tools return verbatim passages with source URLs, cost and stop reasons, and errors arrive as codes with instructions. The project is three days old and alpha, it depends on TypeSafe's paid Jev API, which is in early access with no free tier found, and it does not block crawls of private or local addresses.

Scrapewise

Agents get 14 priced MCP tools by default, each with a free quote_only price check and idempotency keys, read-only or full-access keys, and public euro prices from €0.15 per 1,000 plain pages. There is no status page, SLA or vulnerability disclosure policy, and the terms claim ownership of all scraped data under a perpetual licence.

jevcrawl

1. Phrase goal as one specific question and start with max_pages between 10 and 30
2. Check stats.stopped_because and stats.degraded, and read uncertain before reporting that something is missing
3. Use scrape without a goal when you need a whole page, since it makes no paid Jev call and needs no key
4. Handle {"error": code, "message": ...} results, for example auth_error means TYPESAFE_API_KEY is missing or rejected
5. Pages in stats.needs_js_urls were not read. Tell the user rather than guessing their content

Scrapewise

1. Call each paid tool with quote_only: true first, then repeat with the same idempotency_key so a retry is charged once
2. Connect to https://mcp.scrapewise.ai/mcp?toolsets=outcome,admin to build scrapers; a misspelt toolsets is ignored silently and leaves the 14 data tools
3. Use an LLM_READ key unless the task needs paid tools or writes, since LLM_FULL has no per-tool consent
4. Treat 402 WALLET_INSUFFICIENT_BALANCE as final and ask the user to top up; on 429 wait for Retry-After, and on 429 cost_exceeded wait for the next hour
5. Scrape many addresses with one scrapewise_run_scraper_url_list call (up to 5,000) and retry with the same name, never one scraper per address