What it does
The server adds Google search, webpage content extraction and screenshot capture to Claude. Its tools are search_google, which runs a search and extracts the results, visit_page, which extracts a page's content and can optionally take a screenshot, and take_screenshot, which captures the current page. It also keeps a research session of search queries, visited pages, extracted content, screenshots and timestamps, and exposes screenshots as MCP resources.
It suits chats where Claude needs current information from the web. A bundled agentic-research prompt guides broad searches, favours authoritative sources, refines the direction as findings arrive, checks in with the user and cites URLs. Without that prompt, naming trusted sources in the request, such as a specific news outlet, is suggested as a way to improve results.
Setup requires cloning the repository, installing dependencies, building the project and adding a server entry to the claude_desktop_config.json file in Claude Desktop. Prompts and captured screenshots are then reached through the Paperclip icon in the chat input. Playwright is installed automatically as a dependency. Debugging is done through Claude Desktop's MCP logs.
Limitations and requirements
Requires Node.js 18 or later, the Claude Desktop app, and pnpm or npm. The configuration entry must point to the local path of the cloned repository. Playwright is installed automatically. The publisher lists macOS as a verified platform and Linux as not verified. The repository is a fork of an original mcp-webresearch project.
Safety
Our team has not reviewed this resource. Only automated rules checked it; resources with high-risk signals are not published. These are signals, not a guarantee that installing it is safe. Try it in a sandbox project first.
- Runs code or changes an assistant's behaviour
Sources
- Cline MCP Marketplace first seen 13 Nov 2026 · last seen 17 Nov 2026 · its page on the source
26
As shown on cline.bot when we last read it