home / mcp / webpeel mcp server
Provides MCP endpoints for fetching web pages as clean markdown, searching the web, batch fetching, and crawling with optional stealth and rendering features.
Configuration
View docs{
"mcpServers": {
"jakeliume-webpeel": {
"command": "npx",
"args": [
"-y",
"webpeel",
"mcp"
]
}
}
}WebPeel offers an MCP server integration that lets you access its web content extraction capabilities through interoperable clients such as Claude Code, Cursor, VS Code, and Windsurf. With a lightweight, zero-configuration start and a straightforward setup, you can fetch, search, batch fetch, and crawl web pages from your preferred MCP environment while leveraging WebPeel’s smart escalation and markdown output.
To use the WebPeel MCP server with your MCP client, connect using the standard MCP workflow by registering the WebPeel MCP server and then invoking the provided commands from your client. The server exposes functions such as webpeel_fetch to fetch a URL as clean markdown, webpeel_search to query the web, webpeel_batch to fetch multiple URLs, and webpeel_crawl to crawl a site while respecting robots.txt. Use these functions from your MCP client in the same way you would call other MCP services.
When you configure your MCP client, you will reference the WebPeel MCP server using its stdio command as shown in the examples. The common pattern is to run npx with the WebPeel package and pass mcp as the subcommand. For example, to begin using the MCP server from Claude Code, add the WebPeel MCP server to your client configuration and then call the available endpoints via the client’s MCP interface.
Prerequisites: make sure you have Node.js and npm installed on your system. You do not need to install a separate server to use the MCP integration; you will invoke the MCP endpoints through your MCP client.
Install and prepare your MCP client to talk to WebPeel using the provided command pattern. The WebPeel MCP server is consumed via stdio by MCP clients, typically through npm/npx invocations.
If you are configuring Claude Desktop, Cursor, VS Code, or Windsurf, you will use a ready-made stdio configuration snippet that specifies the command and arguments to start the MCP endpoint.
{
"mcpServers": {
"webpeel": {
"command": "npx",
"args": ["-y", "webpeel", "mcp"]
}
}
}{
"mcpServers": {
"webpeel": {
"command": "npx",
"args": ["-y", "webpeel", "mcp"]
}
}
}{
"mcpServers": {
"webpeel": {
"command": "npx",
"args": ["-y", "webpeel", "mcp"]
}
}
}{
"mcpServers": {
"webpeel": {
"command": "npx",
"args": ["-y", "webpeel", "mcp"]
}
}
}One command to add WebPeel to Claude Code is available. You can run the following to register WebPeel as an MCP server for Claude Code.
claude mcp add webpeel -- npx -y webpeel mcp
You can also place the MCP server configuration in your project's .mcp.json so your team can share access to WebPeel across environments.
{
"mcpServers": {
"webpeel": {
"command": "npx",
"args": ["-y", "webpeel", "mcp"]
}
}
}The MCP interface exposes the following functions you can call from your MCP client: webpeel_fetch to fetch any URL as clean markdown with optional stealth mode, webpeel_search to perform DuckDuckGo-based web searches, webpeel_batch to fetch multiple URLs concurrently, and webpeel_crawl to crawl and extract pages from a site while obeying robots.txt.
Within the MCP server, you can access: webpeel_fetch to retrieve a single URL as markdown, webpeel_search to perform web searches, webpeel_batch for multiple URL fetches, and webpeel_crawl to traverse sites. Each tool is designed to integrate with your LLM workflows to supply clean, structured content.
Fetch any URL and return clean markdown along with metadata and tokens.
Search the web using a defined search endpoint with results suitable for feeding into LLM prompts.
Fetch multiple URLs concurrently through a single MCP invocation.
Crawl a site by following links while respecting robots.txt and rate limits.