MCP NPX Fetch

Instructions

Implementation Reference

• src/Fetcher.ts:61-88 (handler)

  The core handler function for the 'fetch_txt' tool. Fetches the HTML, uses JSDOM to strip scripts/styles and extract body textContent, normalizes whitespace, and returns as plain text content.

  static async txt(requestPayload: RequestPayload) {
    try {
      const response = await this._fetch(requestPayload);
      const html = await response.text();
  
      const dom = new JSDOM(html);
      const document = dom.window.document;
  
      const scripts = document.getElementsByTagName("script");
      const styles = document.getElementsByTagName("style");
      Array.from(scripts).forEach((script) => script.remove());
      Array.from(styles).forEach((style) => style.remove());
  
      const text = document.body.textContent || "";
  
      const normalizedText = text.replace(/\s+/g, " ").trim();
  
      return {
        content: [{ type: "text", text: normalizedText }],
        isError: false,
      };
    } catch (error) {
      return {
        content: [{ type: "text", text: (error as Error).message }],
        isError: true,
      };
    }
  }

• src/index.ts:86-104 (registration)

  Registration of the 'fetch_txt' tool in the ListTools response, including name, description, and input schema.

  {
    name: "fetch_txt",
    description:
      "Fetch a website, return the content as plain text (no HTML)",
    inputSchema: {
      type: "object",
      properties: {
        url: {
          type: "string",
          description: "URL of the website to fetch",
        },
        headers: {
          type: "object",
          description: "Optional headers to include in the request",
        },
      },
      required: ["url"],
    },
  },

• src/index.ts:140-143 (registration)

  Dispatch logic in CallToolRequestSchema handler that routes 'fetch_txt' calls to Fetcher.txt.

  if (request.params.name === "fetch_txt") {
    const fetchResult = await Fetcher.txt(validatedArgs);
    return fetchResult;
  }

• src/types.ts:1-8 (schema)

  Zod schema for request payload (url required, headers optional) used to validate and type inputs for all fetch tools, including fetch_txt.

  import { z } from "zod";
  
  export const RequestPayloadSchema = z.object({
    url: z.string().url(),
    headers: z.record(z.string()).optional(),
  });
  
  export type RequestPayload = z.infer<typeof RequestPayloadSchema>;

• src/Fetcher.ts:6-30 (helper)

  Private helper method for performing the actual HTTP fetch with custom User-Agent and error handling, used by all Fetcher methods including txt.

  private static async _fetch({
    url,
    headers,
  }: RequestPayload): Promise<Response> {
    try {
      const response = await fetch(url, {
        headers: {
          "User-Agent":
            "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36",
          ...headers,
        },
      });
  
      if (!response.ok) {
        throw new Error(`HTTP error: ${response.status}`);
      }
      return response;
    } catch (e: unknown) {
      if (e instanceof Error) {
        throw new Error(`Failed to fetch ${url}: ${e.message}`);
      } else {
        throw new Error(`Failed to fetch ${url}: Unknown error`);
      }
    }
  }