Shark-no-Kari FAQs

Question 1

What is Shark-no-Kari?

Accepted Answer

Shark-no-Kari is a remote MCP (Model Context Protocol) server designed to supercharge Claude's web scraping capabilities. It provides advanced tools for anti-bot evasion, headless browsing, and efficient content extraction that Claude's native `web_fetch` often struggles with.

Question 2

How does Shark-no-Kari help Claude bypass tough websites?

Accepted Answer

It leverages a real headless browser to render JavaScript and bypass protections like Cloudflare. It also uses stealth HTTP fetching with anti-bot headers to access sites that typically block automated requests, including GitHub blob URLs and dynamic pages.

Question 3

What specific content can Shark-no-Kari extract?

Accepted Answer

Shark-no-Kari offers multiple tools: `fetch_page` for stealth HTTP, `stealth_fetch_page` for headless browsing, `extract_elements` for structured data via CSS selectors, and `get_youtube_transcript` for retrieving video captions with language fallback.

Question 4

Is Shark-no-Kari difficult to deploy?

Accepted Answer

No, it's designed for easy deployment. You can quickly set it up on any VPS using Docker and Docker Compose. Once deployed, you simply add its URL as a custom connector in claude.ai settings.

Question 5

How does Shark-no-Kari ensure security and prevent blocks?

Accepted Answer

It uses Caddy as a reverse proxy for auto HTTPS and implements IP allowlisting to restrict access to Anthropic's outbound IP ranges, ensuring only Claude's servers can reach it. For anti-bot evasion, it uses Scrapling's stealth headers and a real headless browser with anti-bot techniques.

Shark-no-Kari

Shark-no-Kari

주요 기능

사용 사례

주요 기능

사용 사례