Skip to content

Quickstart ​

Install ​

bash
npx -y @sylphx/lookout

For Claude Code:

bash
claude mcp add lookout -- npx -y @sylphx/lookout

Then ask for a search or a page. Read the URL, the excerpt span, the warnings, and the gaps before relying on the answer.

What runs ​

There is no profile switch. The tool you name is the work that runs.

CallWhat it does
web_searchQueries DuckDuckGo HTML, Wikipedia OpenSearch, the npm registry, and Hacker News Algolia. It does not fetch those pages.
web_fetchOne HTTP GET. Default 1,500,000 bytes, 20 seconds, at most 5 redirects.
web_extractTitle, metadata, tables, and spans from HTML you pass, or from a URL it fetches.
web_researchSeparate tool. Search, then fetch and extract the top pages (default 3, maximum 6). Search does not call it.

useCache defaults on for search and fetch. A cache hit replays an earlier response. The first call still needs the network. Fetch reads the HTTP body and does not run JavaScript. Private addresses are rejected.

MIT licensed · search and fetch, no API key · these docs load no remote assets