Websearch FAQs

Question 1

What features does Websearch offer for content extraction?

Accepted Answer

Websearch provides dedicated tools for web search, fetching specific URLs, and a combined search-and-fetch operation. You can configure parameters like the maximum number of characters, whether to extract metadata, and the Markdown heading style for fine-tuned content processing.

Question 2

How much can Websearch reduce LLM token usage?

Accepted Answer

Websearch achieves an average token reduction of approximately 60% by converting verbose HTML into concise Markdown. Benchmark results show savings ranging from 44% to 77% on various real-world web pages, directly impacting LLM input costs.

Question 3

What is Websearch and how does it benefit LLMs?

Accepted Answer

Websearch is a powerful tool for web search and content extraction from web pages. It converts fetched HTML into clean Markdown, which significantly reduces the token count (average ~60%) for LLMs, leading to lower operational costs and more efficient processing of web content.

Question 4

Is the HTML-to-Markdown conversion performed efficiently?

Accepted Answer

Yes, Websearch utilizes a high-performance, Rust-based `html-to-markdown` converter, which processes content at approximately 200 MB/s. This ensures rapid and efficient transformation of web pages into clean, LLM-optimized Markdown.

Question 5

Does Websearch require API keys for its web search capabilities?

Accepted Answer

No, Websearch leverages DuckDuckGo for its web search functionality, meaning you do not need any API keys to perform searches. This simplifies setup and reduces dependencies.

Websearch

Websearch

主要功能

使用案例

主要功能

使用案例