This project provides an MCP (Model Context Protocol) server to integrate real-time web research capabilities into Claude Desktop. It allows users to easily bring current information from the web into their AI conversations, enhancing research depth and accuracy for users engaged in detailed topic exploration.

How It Works

The server leverages Node.js and Playwright to interact with web pages. It exposes tools for Google searches, webpage content extraction, and screenshot capture. These tools are integrated with Claude Desktop via a configuration file, enabling the AI to proactively search the web, visit pages, and extract information as part of a research session.

Quick Start & Requirements

• Install Node.js >= 18.
• Add the provided JSON configuration to ~/Library/Application Support/Claude/claude_desktop_config.json.
• Usage is initiated by prompting Claude Desktop, optionally using the provided agentic-research prompt.
• Official documentation and prompt examples are available within the repository.

Highlighted Details

• Integrates directly with Claude Desktop for seamless AI-driven web research.
• Supports automated web page screenshot capture.
• Tracks research sessions, including queries, visited pages, and extracted content.
• Offers an agentic-research prompt for guided, iterative web exploration.

Maintenance & Community

• Developed by mzxrai.
• No specific community channels (Discord/Slack) or roadmap are mentioned in the README.

Licensing & Compatibility

• MIT License.
• Permissive license suitable for commercial use and integration with closed-source applications.

Limitations & Caveats

The project is explicitly described as "very much pre-alpha code" and users should expect bugs. Compatibility is verified for macOS only, with Linux support not yet confirmed.