#!/usr/bin/env node
/**
* ORZ MCP Server - Web Search & Fetch (stdio / Node.js)
*
* 本地运行的 MCP 服务器，通过 stdio 与客户端通信。
* 提供 web_search 和 web_fetch 两个工具。
*
* 特性:
* - 支持 HTTP/HTTPS 代理（命令行参数 --proxy 或环境变量兜底）
* 方便国内用户通过 VPN 访问 Brave 等海外搜索引擎
* - 同时查询 Brave、搜狗、DuckDuckGo，合并去重
* - 抓取网页内容，可选简化为 Markdown
*
* 使用:
* npx -y orz-mcp
* npx -y orz-mcp --proxy http://127.0.0.1:7890
*
* MCP 客户端配置:
* {
* "mcpServers": {
* "orz": {
* "command": "npx",
* "args": ["-y", "orz-mcp", "--proxy", "http://127.0.0.1:7890"]
* }
* }
* }
*/
import { Server } from "@modelcontextprotocol/sdk/server/index.js";
import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js";
import {
CallToolRequestSchema,
ListToolsRequestSchema,
} from "@modelcontextprotocol/sdk/types.js";
import TurndownService from "turndown";
import { ProxyAgent, fetch as undiciFetch } from "undici";
// ============================================================================
// 命令行参数解析
// ============================================================================
/**
* 解析 --proxy <url> 参数
* 优先级: --proxy 命令行参数 > 环境变量 HTTPS_PROXY / HTTP_PROXY / ALL_PROXY
*/
function parseArgs() {
const args = process.argv.slice(2);
let proxy = "";
for (let i = 0; i < args.length; i++) {
if (args[i] === "--proxy" && i + 1 < args.length) {
proxy = args[i + 1];
i++;
} else if (args[i].startsWith("--proxy=")) {
proxy = args[i].slice("--proxy=".length);
} else if (args[i] === "--help" || args[i] === "-h") {
console.error(`
ORZ MCP Server - Web Search & Fetch
Usage:
npx -y orz-mcp [options]
Options:
--proxy <url> HTTP/HTTPS proxy URL (e.g. http://127.0.0.1:7890)
-h, --help Show this help message
Environment variables (used as fallback if --proxy is not set):
HTTPS_PROXY, HTTP_PROXY, ALL_PROXY
Examples:
npx -y orz-mcp --proxy http://127.0.0.1:7890
HTTPS_PROXY=http://127.0.0.1:7890 npx -y orz-mcp
`);
process.exit(0);
}
}
return { proxy };
}
// ============================================================================
// 代理配置
// ============================================================================
const cliArgs = parseArgs();
const PROXY_URL =
cliArgs.proxy ||
process.env.HTTPS_PROXY ||
process.env.https_proxy ||
process.env.HTTP_PROXY ||
process.env.http_proxy ||
process.env.ALL_PROXY ||
process.env.all_proxy ||
"";
/** 如果配置了代理，创建 undici ProxyAgent 作为 dispatcher */
const proxyDispatcher = PROXY_URL
? new ProxyAgent(PROXY_URL)
: undefined;
if (PROXY_URL) {
console.error(`[orz] proxy: ${PROXY_URL}`);
}
// ============================================================================
// 带代理支持的 fetch 封装
// ============================================================================
/**
* 带代理的 fetch 封装
* 使用 undici.fetch + ProxyAgent dispatcher 实现代理支持
*/
function proxyFetch(url, options = {}) {
const fetchOptions = { ...options };
if (proxyDispatcher) {
fetchOptions.dispatcher = proxyDispatcher;
}
return undiciFetch(url, fetchOptions);
}
// ============================================================================
// 常量与配置
// ============================================================================
const USER_AGENTS = [
"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36",
"Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36",
"Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36",
"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/17.2 Safari/605.1.15",
];
function getRandomUA() {
return USER_AGENTS[Math.floor(Math.random() * USER_AGENTS.length)];
}
function getBrowserHeaders() {
return {
"User-Agent": getRandomUA(),
Accept:
"text/html,application/xhtml+xml,application/xml;q=0.9,image/avif,image/webp,*/*;q=0.8",
"Accept-Language": "en-US,en;q=0.9,zh-CN;q=0.8,zh;q=0.7",
"Accept-Encoding": "gzip, deflate, br",
"Cache-Control": "no-cache",
Pragma: "no-cache",
"Sec-Fetch-Dest": "document",
"Sec-Fetch-Mode": "navigate",
"Sec-Fetch-Site": "none",
"Sec-Fetch-User": "?1",
"Upgrade-Insecure-Requests": "1",
};
}
// ============================================================================
// HTML 工具函数
// ============================================================================
function decodeHtmlEntities(text) {
return text
.replace(/&/g, "&")
.replace(/</g, "<")
.replace(/>/g, ">")
.replace(/"/g, '"')
.replace(/'/g, "'")
.replace(/'/g, "'")
.replace(/'/g, "'")
.replace(/&#(\d+);/g, (_, code) => String.fromCharCode(Number(code)))
.replace(/&#x([0-9a-fA-F]+);/g, (_, hex) =>
String.fromCharCode(parseInt(hex, 16))
);
}
function stripHtml(html) {
return decodeHtmlEntities(html.replace(/<[^>]+>/g, "")).trim();
}
// ============================================================================
// 广告过滤
// ============================================================================
function isAdUrl(url) {
const adPatterns = [
/googleads\./i,
/doubleclick\./i,
/googlesyndication\./i,
/googleadservices\./i,
/adclick\./i,
/adsense\./i,
/adservice\./i,
/adserver\./i,
/clickserve\./i,
/clicktrack\./i,
/baidu\.com\/aclick/i,
/pos\.baidu\.com/i,
/cpro\.baidu\.com/i,
/e\.baidu\.com/i,
/bingads\./i,
/microsoftadvertising\./i,
/ad_provider=/i,
/ad_domain=/i,
/\/ads?\//i,
/\/advert/i,
/\/sponsor/i,
/\/promo\//i,
/\/click\?/i,
/\/aclk\?/i,
/\/pagead\//i,
];
return adPatterns.some((p) => p.test(url));
}
// ============================================================================
// 搜索引擎解析
// ============================================================================
/** Brave Search */
function parseBrave(html) {
const results = [];
const blocks = html.split('data-type="web"');
for (let i = 1; i < blocks.length; i++) {
const block = blocks[i].substring(0, 5000);
const urlMatch = block.match(
/<a[^>]+href="(https?:\/\/(?!search\.brave\.com|brave\.com)[^"]+)"/
);
if (!urlMatch) continue;
const url = decodeHtmlEntities(urlMatch[1]);
const aTagMatch = block.match(
/<a[^>]+href="(https?:\/\/(?!search\.brave\.com|brave\.com)[^"]+)"[^>]*>([\s\S]*?)<\/a>/
);
const title = aTagMatch ? stripHtml(aTagMatch[2]) : "";
let summary = "";
const descMatch = block.match(
/class="[^"]*snippet-description[^"]*"[^>]*>([\s\S]*?)<\/(?:p|div|span)>/
);
if (descMatch) {
summary = stripHtml(descMatch[1]);
}
if (!summary) {
const genericMatch = block.match(
/class="[^"]*generic-snippet[^"]*"[^>]*>([\s\S]*?)<\/div>/
);
if (genericMatch) summary = stripHtml(genericMatch[1]);
}
if (title && title.length > 1 && url) {
results.push({ url, title, summary });
}
}
return results;
}
/** 搜狗搜索 */
function parseSogou(html) {
const results = [];
const blocks = html.split('class="vrwrap"');
for (let i = 1; i < blocks.length; i++) {
const block = blocks[i].substring(0, 5000);
const h3Match = block.match(/<h3[^>]*>([\s\S]*?)<\/h3>/);
if (!h3Match) continue;
const linkMatch = h3Match[1].match(
/<a[^>]+href="([^"]+)"[^>]*>([\s\S]*?)<\/a>/
);
if (!linkMatch) continue;
let url = decodeHtmlEntities(linkMatch[1]);
const title = stripHtml(linkMatch[2]);
if (!title) continue;
if (url.startsWith("/link?")) {
url = "https://www.sogou.com" + url;
}
let summary = "";
const summaryPatterns = [
/class="[^"]*text-layout[^"]*"[^>]*>([\s\S]*?)<\/div>/,
/class="[^"]*summary[^"]*"[^>]*>([\s\S]*?)<\/div>/,
/class="[^"]*str[-_]text[^"]*"[^>]*>([\s\S]*?)<\/(?:p|div)>/,
];
for (const pat of summaryPatterns) {
const m = block.match(pat);
if (m) {
const text = stripHtml(m[1]);
if (text.length > 10) {
summary = text;
break;
}
}
}
results.push({ url, title, summary });
}
return results;
}
/** DuckDuckGo */
function parseDuckDuckGo(html) {
const results = [];
if (
html.includes("anomaly-modal") ||
html.includes("Please complete the following challenge")
) {
console.error("[DuckDuckGo] Got captcha page, skipping");
return results;
}
const linkRegex =
/<a[^>]*class="[^"]*result__a[^"]*"[^>]*href="([^"]*)"[^>]*>([\s\S]*?)<\/a>/gi;
const snippetRegex =
/<a[^>]*class="[^"]*result__snippet[^"]*"[^>]*>([\s\S]*?)<\/a>/gi;
const links = [];
let linkMatch;
while ((linkMatch = linkRegex.exec(html)) !== null) {
links.push({ rawUrl: linkMatch[1], title: stripHtml(linkMatch[2]) });
}
const snippets = [];
let snippetMatch;
while ((snippetMatch = snippetRegex.exec(html)) !== null) {
snippets.push(stripHtml(snippetMatch[1]));
}
for (let i = 0; i < links.length; i++) {
let { rawUrl, title } = links[i];
rawUrl = decodeHtmlEntities(rawUrl);
let url = rawUrl;
if (rawUrl.includes("uddg=")) {
const uddgParam = rawUrl.split("uddg=")[1]?.split("&")[0] ?? "";
const decoded = decodeURIComponent(uddgParam);
if (decoded) url = decoded;
} else if (rawUrl.startsWith("//")) {
url = "https:" + rawUrl;
}
if (url.includes("duckduckgo.com/y.js") || isAdUrl(url)) continue;
const summary = i < snippets.length ? snippets[i] : "";
if (title && url) results.push({ url, title, summary });
}
return results;
}
// ============================================================================
// 搜索引擎请求
// ============================================================================
async function searchBrave(query) {
try {
const url = `https://search.brave.com/search?q=${encodeURIComponent(query)}`;
const resp = await proxyFetch(url, {
headers: getBrowserHeaders(),
signal: AbortSignal.timeout(10000),
});
if (!resp.ok) return [];
return parseBrave(await resp.text());
} catch (e) {
console.error("[Brave] search error:", e.message);
return [];
}
}
async function searchSogou(query) {
try {
const url = `https://www.sogou.com/web?query=${encodeURIComponent(query)}`;
const resp = await proxyFetch(url, {
headers: getBrowserHeaders(),
signal: AbortSignal.timeout(10000),
});
if (!resp.ok) return [];
return parseSogou(await resp.text());
} catch (e) {
console.error("[Sogou] search error:", e.message);
return [];
}
}
async function searchDuckDuckGo(query) {
try {
const url = `https://html.duckduckgo.com/html/?q=${encodeURIComponent(query)}`;
const resp = await proxyFetch(url, {
headers: getBrowserHeaders(),
signal: AbortSignal.timeout(10000),
});
if (!resp.ok) return [];
return parseDuckDuckGo(await resp.text());
} catch (e) {
console.error("[DuckDuckGo] search error:", e.message);
return [];
}
}
// ============================================================================
// 去重合并
// ============================================================================
function normalizeUrl(url) {
try {
const parsed = new URL(url);
const host = parsed.hostname.replace(/^www\./, "");
const path = parsed.pathname.replace(/\/+$/, "");
const trackingParams = [
"utm_source", "utm_medium", "utm_campaign", "utm_term", "utm_content",
"ref", "fbclid", "gclid", "msclkid", "spm", "from",
];
const params = new URLSearchParams(parsed.search);
for (const tp of trackingParams) params.delete(tp);
const search = params.toString() ? `?${params.toString()}` : "";
return `${host}${path}${search}`.toLowerCase();
} catch {
return url.toLowerCase();
}
}
function mergeAndDeduplicate(allResults, maxResults) {
const seen = new Set();
const merged = [];
const maxLen = Math.max(...allResults.map((r) => r.length), 0);
for (let idx = 0; idx < maxLen; idx++) {
for (const engineResults of allResults) {
if (idx >= engineResults.length) continue;
const item = engineResults[idx];
if (isAdUrl(item.url)) continue;
const normalized = normalizeUrl(item.url);
if (seen.has(normalized)) continue;
seen.add(normalized);
merged.push(item);
if (merged.length >= maxResults) return merged;
}
}
return merged;
}
// ============================================================================
// web_search
// ============================================================================
async function webSearch(query, numResults = 8) {
console.error(`[web_search] query="${query}", numResults=${numResults}`);
const [brave, sogou, ddg] = await Promise.allSettled([
searchBrave(query),
searchSogou(query),
searchDuckDuckGo(query),
]);
const allResults = [];
const engines = [
["Brave", brave],
["Sogou", sogou],
["DuckDuckGo", ddg],
];
for (const [name, result] of engines) {
if (result.status === "fulfilled") {
allResults.push(result.value);
console.error(`[web_search] ${name}: ${result.value.length} results`);
} else {
console.error(`[web_search] ${name}: failed - ${result.reason}`);
}
}
return mergeAndDeduplicate(allResults, numResults);
}
// ============================================================================
// web_fetch
// ============================================================================
function removeUselessTags(html) {
const tags = [
"script", "style", "iframe", "noscript", "svg",
"object", "embed", "applet", "link", "meta",
"head", "nav", "footer", "aside",
];
let cleaned = html;
for (const tag of tags) {
cleaned = cleaned.replace(
new RegExp(`<${tag}[\\s\\S]*?(?:<\\/${tag}>|\\/>)`, "gi"),
""
);
}
return cleaned.replace(/<!--[\s\S]*?-->/g, "");
}
function htmlToMarkdown(html) {
const td = new TurndownService({
headingStyle: "atx",
codeBlockStyle: "fenced",
bulletListMarker: "-",
});
td.remove(["script", "style", "iframe", "noscript", "svg", "nav", "footer"]);
try {
let md = td.turndown(html);
md = md.replace(/\n{3,}/g, "\n\n").replace(/[ \t]+$/gm, "");
return md.trim();
} catch {
return html.replace(/<[^>]+>/g, " ").replace(/\s+/g, " ").trim();
}
}
function extractMainContent(html) {
for (const [tag] of [["main"], ["article"]]) {
const m = html.match(new RegExp(`<${tag}[^>]*>([\\s\\S]*?)<\\/${tag}>`, "i"));
if (m) return m[1];
}
const contentMatch = html.match(
/<div[^>]*id=["']content["'][^>]*>([\s\S]*?)<\/div>/i
);
if (contentMatch) return contentMatch[1];
const bodyMatch = html.match(/<body[^>]*>([\s\S]*?)<\/body>/i);
if (bodyMatch) return bodyMatch[1];
return html;
}
async function webFetch(url, maxCharSize = 50000, simplify = true) {
console.error(
`[web_fetch] url="${url}", maxCharSize=${maxCharSize}, simplify=${simplify}`
);
const controller = new AbortController();
const timeoutId = setTimeout(() => controller.abort(), 10000);
try {
const resp = await proxyFetch(url, {
headers: getBrowserHeaders(),
signal: controller.signal,
redirect: "follow",
});
clearTimeout(timeoutId);
if (!resp.ok) throw new Error(`HTTP ${resp.status}: ${resp.statusText}`);
const contentType = resp.headers.get("content-type") ?? "";
if (!contentType.includes("html")) {
return (await resp.text()).substring(0, maxCharSize);
}
let html = await resp.text();
if (simplify) {
html = removeUselessTags(html);
const main = extractMainContent(html);
return htmlToMarkdown(main).substring(0, maxCharSize);
}
return html.substring(0, maxCharSize);
} catch (e) {
clearTimeout(timeoutId);
if (e.name === "AbortError") {
throw new Error(`Timeout: Failed to fetch "${url}" within 10 seconds.`);
}
throw e;
}
}
// ============================================================================
// MCP 工具定义
// ============================================================================
const TOOLS = [
{
name: "web_search",
description:
"Search the web using multiple search engines (Brave, Sogou, DuckDuckGo) simultaneously. " +
"Results are deduplicated and ads are filtered out. " +
"Returns an array of search results with url, title, and summary.",
inputSchema: {
type: "object",
properties: {
query: {
type: "string",
description: "Search keywords separated by spaces",
},
num_results: {
type: "number",
description: "Number of results to return (default: 8)",
default: 8,
},
},
required: ["query"],
},
},
{
name: "web_fetch",
description:
"Fetch a web page and return its content. " +
"When simplify is enabled (default), removes useless HTML tags and converts to Markdown. " +
"Has a 10-second timeout.",
inputSchema: {
type: "object",
properties: {
url: {
type: "string",
description: "The URL to fetch",
},
max_char_size: {
type: "number",
description:
"Maximum character size of the returned content (default: 50000)",
default: 50000,
},
simplify: {
type: "boolean",
description:
"Whether to simplify the content by removing useless tags and converting to Markdown (default: true)",
default: true,
},
},
required: ["url"],
},
},
];
// ============================================================================
// 工具调用处理
// ============================================================================
async function handleToolCall(name, args) {
switch (name) {
case "web_search": {
const query = args.query;
const numResults = args.num_results ?? 8;
if (!query || query.trim() === "") {
return {
content: [{ type: "text", text: "Error: query is required." }],
isError: true,
};
}
try {
const results = await webSearch(query, numResults);
return {
content: [{ type: "text", text: JSON.stringify(results, null, 2) }],
};
} catch (e) {
return {
content: [{ type: "text", text: `Error: ${e.message}` }],
isError: true,
};
}
}
case "web_fetch": {
const url = args.url;
const maxCharSize = args.max_char_size ?? 50000;
const simplify = args.simplify ?? true;
if (!url || url.trim() === "") {
return {
content: [{ type: "text", text: "Error: url is required." }],
isError: true,
};
}
try {
const content = await webFetch(url, maxCharSize, simplify);
return { content: [{ type: "text", text: content }] };
} catch (e) {
return {
content: [{ type: "text", text: `Error: ${e.message}` }],
isError: true,
};
}
}
default:
return {
content: [{ type: "text", text: `Unknown tool: ${name}` }],
isError: true,
};
}
}
// ============================================================================
// MCP Server 启动 (stdio)
// ============================================================================
const server = new Server(
{ name: "orz", version: "1.0.0" },
{ capabilities: { tools: {} } }
);
server.setRequestHandler(ListToolsRequestSchema, () => ({ tools: TOOLS }));
server.setRequestHandler(CallToolRequestSchema, (request) =>
handleToolCall(request.params.name, request.params.arguments ?? {})
);
const transport = new StdioServerTransport();
await server.connect(transport);
console.error("[orz] MCP server running on stdio");
