Raw Web Search FAQs

Question 1

Can Raw Web Search replace WebFetch for URL content extraction?

Accepted Answer

Yes, its `extract` tool is designed as a direct replacement for WebFetch, offering a near 100% success rate for robust content extraction from specific URLs, significantly outperforming WebFetch's reported ~50% success.

Question 2

What is Raw Web Search for Claude Code?

Accepted Answer

Raw Web Search is an experimental MCP search server designed for Claude Code that retrieves raw, verbatim web content, code examples, and API documentation directly from sources, instead of AI-rewritten summaries.

Question 3

Does Raw Web Search have any external dependencies?

Accepted Answer

Yes, it requires Python 3.10+ and Docker to run its SearXNG search backend. Installing Chromium for Playwright is optional but highly recommended to achieve the highest content extraction success rates.

Question 4

How does Raw Web Search compare to Claude's built-in WebSearch?

Accepted Answer

Raw Web Search is superior for code examples, API docs (due to verbatim output), batch searching (auto-approvable), and direct URL extraction. WebSearch is faster and generally better for synthesized overviews and broad factual knowledge.

Question 5

What are the primary use cases for Raw Web Search?

Accepted Answer

It's best used when you need exact source text like code snippets, API signatures, or precise quotes. It's also ideal for scenarios requiring frictionless, auto-approved batch web searching without per-call user approval.

Raw Web Search

Raw Web Search

主な機能

ユースケース

主な機能

ユースケース