mcp -fetch-typescript

The mcp-server-fetch-typescript MCP Server provides web content fetching and conversion capabilities. It exposes tools to retrieve raw text, render fully-rendered HTML, convert pages to Markdown, and extract clean Markdown summaries focused on the main content. This makes it suitable for tasks ranging from simple data extraction to archiving and documentation generation for modern web applications that rely on client-side rendering.

Available tools:

• get_raw_text: Fetches raw text content directly from a URL without browser rendering. Ideal for JSON, XML, CSV, TSV, or plain text files when you need direct access to the source content.
• get_rendered_html: Fetches fully rendered HTML content after executing JavaScript. Uses Playwright for a headless browser rendering, essential for modern SPAs and dynamic pages.
• get_markdown: Converts web content to Markdown while preserving structure, including tables and definition lists. Useful for documentation and knowledge bases.
• get_markdown_summary: Extracts the main content and converts it to Markdown, automatically removing navigation, headers, and footers for clean article or blog post summaries.

To use these tools, configure your MCP client to call the server with the appropriate tool name and a required url parameter. You can then process or store the returned Markdown or HTML as needed for your workflow.

Prerequisites:

• Node.js and npm installed on your system
• Access to npm to install packages (for global or project installation)

Installation options:

1. Global installation (recommended for quick testing):

npm install -g mcp-server-fetch-typescript

2. Project-based installation (recommended for development):

npm install mcp-server-fetch-typescript

Building after cloning (if you intend to run from source):

git clone https://github.com/tatn/mcp-server-fetch-typescript.git
cd mcp-server-fetch-typescript
npm install
npm run build

To run a built server directly (example):

node /path/to/mcp-server-fetch-typescript/build/index.js

If you are using the MCP configuration, you can reference the server via npx as shown in the mcp_config section.

Tips and common issues:

• Ensure you have the necessary dependencies installed, including Playwright, which is used for rendering in get_rendered_html. If you encounter failing browser installs, run the server with sufficient permissions or install browsers manually as instructed by Playwright.
• When using get_rendered_html, be aware of dynamic content that may require specific timing or navigation steps on the page; the server handles standard rendering but complex interactions may need additional handling on the client side.
• If running as a project dependency, keep the package updated to benefit from fixes and new features.
• Environment variables or config tweaks can be added in your MCP client configuration if the server supports them in future releases; currently, the primary integration is via the tool interface and URL parameter.
• For debugging, you can use the inspector tool as shown in the README to step through requests and tool invocations.