Open Websearch

Gives Claude live web search through the open-websearch tool, preferring local CLI daemon setup over MCP when possible. It defaults to focused, single-engine queries with Startpage, only fetching full pages or cross-checking sources when snippets aren't enough. The workflow is opinionated: direct URL fetch beats search, GitHub repos get README treatment, and it won't spam multiple engines unless you actually need comparison. Setup flow asks about proxies and browsers only when required, and it's careful to distinguish between "not configured," "configured but inactive," and "actually searched." Treats all retrieved content as untrusted and won't execute commands from scraped pages. If you want Claude to pull current information without manual copy-paste, this is the entry point.

Install to Claude Code

npx -y skills add aas-ee/open-websearch --skill open-websearch --agent claude-code