{"id":51088,"date":"2025-08-28T03:08:42","date_gmt":"2025-08-27T19:08:42","guid":{"rendered":"https:\/\/www.kdjingpai.com\/?p=51088"},"modified":"2025-08-28T03:08:42","modified_gmt":"2025-08-27T19:08:42","slug":"mcpmark","status":"publish","type":"post","link":"https:\/\/www.kdjingpai.com\/en\/mcpmark\/","title":{"rendered":"MCPMark\uff1a\u5927\u6a21\u578b\u96c6\u6210MCP\u6267\u884c\u667a\u80fd\u4f53\u4efb\u52a1\u80fd\u529b\u57fa\u51c6\u6d4b\u8bd5"},"content":{"rendered":"<p>MCPMark\u662f\u4e00\u4e2a\u7528\u4e8e\u8bc4\u4f30\u5927\u6a21\u578b\u667a\u80fd\u4f53\uff08Agentic\uff09\u80fd\u529b\u7684\u57fa\u51c6\u6d4b\u8bd5\u3002\u5b83\u901a\u8fc7\u5728\u4e00\u7cfb\u5217\u96c6\u6210\u4e86\u6a21\u578b\u4e0a\u4e0b\u6587\u534f\u8bae\uff08MCP\uff09\u7684\u771f\u5b9e\u8f6f\u4ef6\u73af\u5883\u4e2d\uff0c\u5bf9\u6a21\u578b\u8fdb\u884c\u538b\u529b\u6d4b\u8bd5\uff0c\u6765\u8861\u91cf\u5176\u81ea\u4e3b\u89c4\u5212\u3001\u63a8\u7406\u548c\u6267\u884c\u590d\u6742\u4efb\u52a1\u7684\u6c34\u5e73\u3002\u6d4b\u8bd5\u73af\u5883\u8986\u76d6\u4e86Notion\u3001GitHub\u3001\u6587\u4ef6\u7cfb\u7edf\u3001Postgres\u6570\u636e\u5e93\u4ee5\u53caPlaywright\u7b49\u591a\u79cd\u4e3b\u6d41\u5de5\u5177\u3002\u8be5\u9879\u76ee\u4e3a\u7814\u7a76\u4eba\u5458\u548c\u5de5\u7a0b\u5e08\u8bbe\u8ba1\uff0c\u901a\u8fc7\u5b89\u5168\u7684\u6c99\u76d2\u673a\u5236\u3001\u53ef\u590d\u73b0\u7684\u81ea\u52a8\u5316\u4efb\u52a1\u548c\u7edf\u4e00\u7684\u8bc4\u4f30\u6307\u6807\uff0c\u63d0\u4f9b\u4e86\u4e00\u4e2a\u5ba2\u89c2\u3001\u53ef\u9760\u7684\u8bc4\u4f30\u5e73\u53f0\u3002<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-51101\" src=\"https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/08\/1f51f116033f53e.png\" alt=\"\" width=\"1298\" height=\"805\" srcset=\"https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/08\/1f51f116033f53e.png 1298w, https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/08\/1f51f116033f53e-18x12.png 18w\" sizes=\"auto, (max-width: 1298px) 100vw, 1298px\" \/><\/p>\n<p>&nbsp;<\/p>\n<h2>\u529f\u80fd\u5217\u8868<\/h2>\n<ul>\n<li><strong>\u591a\u6837\u7684\u6d4b\u8bd5\u73af\u5883<\/strong>: \u652f\u6301\u5728\u516d\u79cd\u771f\u5b9e\u4e14\u590d\u6742\u7684\u8f6f\u4ef6\u73af\u5883\u4e2d\u8fdb\u884c\u6d4b\u8bd5\uff0c\u5305\u62ec<code><a href=\"https:\/\/www.kdjingpai.com\/en\/notion-ai-2\/\">Notion<\/a><\/code>\u3001<code>GitHub<\/code>\u3001<code>Filesystem<\/code>\u3001<code>Postgres<\/code>\u3001<code>Playwright<\/code>\u548c<code>Playwright-WebArena<\/code>\u3002<\/li>\n<li><strong>\u81ea\u52a8\u5316\u7684\u4efb\u52a1\u9a8c\u8bc1<\/strong>: \u6bcf\u4e2a\u6d4b\u8bd5\u4efb\u52a1\u90fd\u914d\u6709\u4e25\u683c\u7684\u81ea\u52a8\u5316\u9a8c\u8bc1\u811a\u672c\uff0c\u80fd\u591f\u5ba2\u89c2\u3001\u53ef\u590d\u73b0\u5730\u5bf9\u4efb\u52a1\u5b8c\u6210\u60c5\u51b5\u8fdb\u884c\u8bc4\u4f30\u3002<\/li>\n<li><strong>\u5b89\u5168\u7684\u6c99\u76d2\u673a\u5236<\/strong>: \u6240\u6709\u4efb\u52a1\u90fd\u5728\u72ec\u7acb\u7684\u6c99\u76d2\u73af\u5883\u4e2d\u8fd0\u884c\uff0c\u4efb\u52a1\u7ed3\u675f\u540e\u73af\u5883\u5373\u88ab\u9500\u6bc1\uff0c\u4e0d\u4f1a\u6cc4\u9732\u6216\u6c61\u67d3\u7528\u6237\u7684\u4e2a\u4eba\u6570\u636e\u3002<\/li>\n<li><strong>\u5931\u8d25\u81ea\u52a8\u7eed\u8dd1<\/strong>: \u5f53\u5b9e\u9a8c\u56e0\u7f51\u7edc\u6ce2\u52a8\u7b49\u7ba1\u9053\u9519\u8bef\uff08Pipeline Error\uff09\u4e2d\u65ad\u540e\uff0c\u91cd\u65b0\u8fd0\u884c\u65f6\u4f1a\u81ea\u52a8\u8df3\u8fc7\u5df2\u5b8c\u6210\u7684\u4efb\u52a1\uff0c\u5e76\u91cd\u8bd5\u4e4b\u524d\u5931\u8d25\u7684\u4efb\u52a1\u3002<\/li>\n<li><strong>\u4e30\u5bcc\u7684\u8bc4\u4f30\u6307\u6807<\/strong>: \u652f\u6301\u751f\u6210\u591a\u79cd\u805a\u5408\u6307\u6807\uff0c\u5305\u62ec<code>pass@1<\/code>\u3001<code>pass@K<\/code>\u3001<code>pass^K<\/code>\u548c<code>avg@K<\/code>\uff0c\u7528\u4e8e\u5168\u9762\u8861\u91cf\u6a21\u578b\u7684\u5355\u6b21\u6210\u529f\u7387\u548c\u591a\u6b21\u5c1d\u8bd5\u7684\u7a33\u5b9a\u6027\u3002<\/li>\n<li><strong>\u7075\u6d3b\u7684\u90e8\u7f72\u9009\u9879<\/strong>: \u652f\u6301\u901a\u8fc7Pip\u5728\u672c\u5730\uff08macOS\u3001Linux\uff09\u5b89\u88c5\uff0c\u4e5f\u63d0\u4f9b\u4e86Docker\u955c\u50cf\uff0c\u65b9\u4fbf\u5feb\u901f\u90e8\u7f72\u548c\u8fd0\u884c\u3002<\/li>\n<\/ul>\n<h2>\u4f7f\u7528\u5e2e\u52a9<\/h2>\n<p>\u4f7f\u7528MCPMark\u8bc4\u4f30\u6a21\u578b\u901a\u5e38\u9075\u5faa\u4ee5\u4e0b\u56db\u4e2a\u6b65\u9aa4\uff1a<\/p>\n<h3><strong>1. \u5b89\u88c5MCPMark<\/strong><\/h3>\n<p>\u4f60\u53ef\u4ee5\u9009\u62e9\u672c\u5730\u5b89\u88c5\u6216\u4f7f\u7528Docker\u3002<br \/>\n<strong>\u672c\u5730\u5b89\u88c5 (Pip):<\/strong><\/p>\n<pre><code># \u4eceGitHub\u514b\u9686\u4ed3\u5e93\r\ngit clone https:\/\/github.com\/eval-sys\/mcpmark.git\r\ncd mcpmark\r\n# \u5b89\u88c5\u4f9d\u8d56\r\npip install -e .\r\n<\/code><\/pre>\n<p><strong>Docker\u5b89\u88c5:<\/strong><\/p>\n<pre><code># \u514b\u9686\u4ed3\u5e93\u540e\uff0c\u76f4\u63a5\u6784\u5efaDocker\u955c\u50cf\r\n.\/build-docker.sh```\r\n\r\n### **2. \u6388\u6743\u670d\u52a1**\r\n\u5982\u679c\u4f60\u9700\u8981\u6d4b\u8bd5GitHub\u6216Notion\u76f8\u5173\u7684\u4efb\u52a1\uff0c\u4f60\u9700\u8981\u5148\u6839\u636e\u5b98\u65b9\u6587\u6863\u8fdb\u884c\u6388\u6743\uff0c\u8ba9MCPMark\u80fd\u591f\u4ee5\u7f16\u7a0b\u65b9\u5f0f\u8bbf\u95ee\u8fd9\u4e9b\u670d\u52a1\u3002\r\n\r\n### **3. \u914d\u7f6e\u73af\u5883\u53d8\u91cf**\r\n\u5728\u9879\u76ee\u6839\u76ee\u5f55\u521b\u5efa\u4e00\u4e2a\u540d\u4e3a&lt;code&gt;.mcp_env&lt;\/code&gt;\u7684\u6587\u4ef6\uff0c\u5e76\u586b\u5165\u4f60\u9700\u8981\u7684\u6a21\u578bAPI\u5bc6\u94a5\u548c\u76f8\u5173\u670d\u52a1\u7684\u6388\u6743\u51ed\u8bc1\u3002\r\n```dotenv\r\n# \u793a\u4f8b\uff1a\u914d\u7f6eOpenAI\u6a21\u578b\r\nOPENAI_BASE_URL=\"https:\/\/api.openai.com\/v1\"\r\nOPENAI_API_KEY=\"sk-...\"\r\n\r\n# \u793a\u4f8b\uff1a\u914d\u7f6eGitHub\r\nGITHUB_TOKENS=\"your_github_token\"\r\nGITHUB_EVAL_ORG=\"your_eval_org\"\r\n\r\n# \u793a\u4f8b\uff1a\u914d\u7f6eNotion\r\nSOURCE_NOTION_API_KEY=\"your_source_notion_api_key\"\r\nEVAL_NOTION_API_KEY=\"your_eval_notion_api_key\"\r\n<\/code><\/pre>\n<h3><strong>4. \u8fd0\u884c\u8bc4\u4f30\u5b9e\u9a8c<\/strong><\/h3>\n<p>\u4f60\u53ef\u4ee5\u6839\u636e\u9700\u6c42\uff0c\u8fd0\u884c\u4e0d\u540c\u8303\u56f4\u7684\u4efb\u52a1\u3002<\/p>\n<pre><code># \u5047\u8bbe\u5b9e\u9a8c\u540d\u4e3a new_exp\uff0c\u6a21\u578b\u4e3a gpt-4.1\uff0c\u73af\u5883\u4e3a notion\uff0c\u8fd0\u884cK\u6b21\r\n\r\n# \u8bc4\u4f30\u8be5\u73af\u5883\u4e0b\u7684\u6240\u6709\u4efb\u52a1\r\npython -m pipeline --exp-name new_exp --mcp notion --tasks all --models <a href=\"https:\/\/www.kdjingpai.com\/en\/gpt-41-guanfangtishi\/\">gpt-4.1<\/a> --k K\r\n\r\n# \u8bc4\u4f30\u4e00\u4e2a\u4efb\u52a1\u7ec4 (\u4f8b\u5982 online_resume)\r\npython -m pipeline --exp-name new_exp --mcp notion --tasks online_resume --models gpt-4.1 --k K\r\n<\/code><\/pre>\n<h3><strong>5. \u67e5\u770b\u4e0e\u805a\u5408\u7ed3\u679c<\/strong><\/h3>\n<p>\u5b9e\u9a8c\u7ed3\u679c\u4f1a\u4ee5JSON\u548cCSV\u683c\u5f0f\u4fdd\u5b58\u5728<code>.\/results\/<\/code>\u76ee\u5f55\u4e2d\u3002\u5982\u679c\u4f60\u7684\u8fd0\u884c\u6b21\u6570K\u5927\u4e8e1\uff0c\u53ef\u4ee5\u8fd0\u884c\u4ee5\u4e0b\u547d\u4ee4\u6765\u751f\u6210\u805a\u5408\u62a5\u544a\u3002<\/p>\n<pre><code>python -m src.aggregators.aggregate_results --exp-name new_exp\r\n<\/code><\/pre>\n<h2>\u5e94\u7528\u573a\u666f<\/h2>\n<ol>\n<li><strong>\u8bc4\u4f30\u6a21\u578b\u7684\u667a\u80fd\u4f53\u80fd\u529b<\/strong><br \/>\n\u7814\u7a76\u673a\u6784\u548c\u5f00\u53d1\u8005\u53ef\u4ee5\u4f7f\u7528\u6b64\u57fa\u51c6\uff0c\u5ba2\u89c2\u8861\u91cf\u4e0d\u540c\u524d\u6cbfAI\u6a21\u578b\u5728\u5904\u7406\u590d\u6742\u5de5\u4f5c\u6d41\u65f6\u7684\u81ea\u4e3b\u89c4\u5212\u3001\u63a8\u7406\u548c\u5de5\u5177\u4f7f\u7528\u80fd\u529b\uff0c\u800c\u4e0d\u4ec5\u4ec5\u662f\u7b80\u5355\u7684API\u8c03\u7528\u3002<\/li>\n<li><strong>AI\u667a\u80fd\u4f53\u56de\u5f52\u6d4b\u8bd5<\/strong><br \/>\n\u5bf9\u4e8e\u5f00\u53d1AI\u667a\u80fd\u4f53\u5e94\u7528\u7684\u56e2\u961f\uff0cMCPMark\u53ef\u4ee5\u4f5c\u4e3a\u4e00\u4e2a\u6807\u51c6\u7684\u56de\u5f52\u6d4b\u8bd5\u96c6\uff0c\u786e\u4fdd\u6a21\u578b\u6216\u5e94\u7528\u7684\u8fed\u4ee3\u66f4\u65b0\u4e0d\u4f1a\u5bfc\u81f4\u5176\u667a\u80fd\u4f53\u80fd\u529b\u7684\u9000\u5316\u3002<\/li>\n<li><strong>\u667a\u80fd\u4f53AI\u7684\u5b66\u672f\u7814\u7a76<\/strong><br \/>\n\u5b66\u8005\u53ef\u4ee5\u5229\u7528\u8fd9\u4e2a\u6807\u51c6\u5316\u7684\u5e73\u53f0\u6765\u53d1\u5e03\u53ef\u590d\u73b0\u7684\u5173\u4e8eAI\u667a\u80fd\u4f53\u80fd\u529b\u7684\u7814\u7a76\u6210\u679c\uff0c\u63a8\u52a8\u6574\u4e2a\u9886\u57df\u7684\u8fdb\u6b65\u3002<\/li>\n<li><strong>\u9a8c\u8bc1\u4e1a\u52a1\u6d41\u7a0b\u7684\u81ea\u4e3b\u5316\u6c34\u5e73<\/strong><br \/>\n\u4f01\u4e1a\u53ef\u4ee5\u5229\u7528MCPMark\u6765\u6d4b\u8bd5AI\u6a21\u578b\u5728\u7279\u5b9a\u4e1a\u52a1\u573a\u666f\uff08\u5982\u4ee3\u7801\u4ed3\u5e93\u7ba1\u7406\u3001\u6570\u636e\u5e93\u64cd\u4f5c\uff09\u4e2d\u80fd\u591f\u8fbe\u5230\u7684\u81ea\u4e3b\u81ea\u52a8\u5316\u6c34\u5e73\u3002<\/li>\n<\/ol>\n<h2>QA<\/h2>\n<ol>\n<li><strong>MCPMark\u5230\u5e95\u662f\u4ec0\u4e48\uff1f<\/strong><br \/>\n\u5b83\u662f\u4e00\u4e2a\u6807\u51c6\u7684\u57fa\u51c6\u6d4b\u8bd5\u5de5\u5177\uff0c\u4e0d\u662f\u4e00\u4e2a\u666e\u901a\u7528\u6237\u4f7f\u7528\u7684AI\u5e94\u7528\u3002\u5b83\u7684\u6838\u5fc3\u76ee\u7684\u662f\u63d0\u4f9b\u4e00\u5957\u53ef\u9760\u7684\u73af\u5883\u548c\u4efb\u52a1\uff0c\u7528\u6765\u79d1\u5b66\u5730\u8bc4\u4f30\u548c\u6bd4\u8f83\u4e0d\u540cAI\u5927\u6a21\u578b\u4f5c\u4e3a\u201c\u667a\u80fd\u4f53\u201d\uff08Agent\uff09\u81ea\u4e3b\u5b8c\u6210\u590d\u6742\u4efb\u52a1\u7684\u80fd\u529b\u3002<\/li>\n<li><strong>\u4ec0\u4e48\u662fMCP\uff08\u6a21\u578b\u4e0a\u4e0b\u6587\u534f\u8bae\uff09\uff1f<\/strong><br \/>\nMCP\uff08Model Context Protocol\uff09\u662f\u4e00\u5957\u6280\u672f\u6807\u51c6\u548c\u534f\u8bae\uff0c\u7528\u4e8e\u89c4\u8303AI\u5927\u6a21\u578b\u4e0e\u5916\u90e8\u5de5\u5177\u548c\u8f6f\u4ef6\u73af\u5883\u4e4b\u95f4\u7684\u4ea4\u4e92\u65b9\u5f0f\u3002MCPMark\u6b63\u662f\u57fa\u4e8e\u8fd9\u5957\u534f\u8bae\u6784\u5efa\u7684\uff0c\u4ee5\u786e\u4fdd\u6a21\u578b\u4e0e\u73af\u5883\u7684\u4e92\u52a8\u662f\u53ef\u63a7\u3001\u53ef\u8861\u91cf\u4e14\u53ef\u590d\u73b0\u7684\u3002<\/li>\n<li><strong>\u8fd0\u884cMCPMark\u6d4b\u8bd5\u662f\u5426\u5b89\u5168\uff1f<\/strong><br \/>\n\u662f\u7684\uff0c\u975e\u5e38\u5b89\u5168\u3002\u5b83\u5728\u4e3a\u6bcf\u6b21\u5b9e\u9a8c\u521b\u5efa\u7684\u3001\u5b8c\u5168\u9694\u79bb\u7684\u6c99\u76d2\u73af\u5883\u4e2d\u8fd0\u884c\u3002\u4efb\u52a1\u4e00\u7ed3\u675f\uff0c\u8fd9\u4e2a\u73af\u5883\u5c31\u4f1a\u88ab\u5f7b\u5e95\u9500\u6bc1\uff0c\u56e0\u6b64\u4e0d\u4f1a\u89e6\u78b0\u6216\u4fee\u6539\u4f60\u672c\u673a\u7684\u4efb\u4f55\u4e2a\u4eba\u6587\u4ef6\u6216\u8d26\u6237\u6570\u636e\u3002<\/li>\n<li><strong>\u4ec0\u4e48\u662fpass@K\u6307\u6807\uff1f<\/strong><br \/>\n<code>pass@K<\/code>\u662f\u8861\u91cf\u6a21\u578b\u53ef\u9760\u6027\u7684\u4e00\u4e2a\u5173\u952e\u6307\u6807\u3002\u5b83\u8868\u793a\u5728K\u6b21\u72ec\u7acb\u7684\u5c1d\u8bd5\u4e2d\uff0c\u6a21\u578b\u81f3\u5c11\u6709\u4e00\u6b21\u6210\u529f\u5b8c\u6210\u4efb\u52a1\u7684\u6982\u7387\u3002\u8fd9\u4e2a\u6307\u6807\u8d8a\u9ad8\uff0c\u8bf4\u660e\u6a21\u578b\u5b8c\u6210\u4efb\u52a1\u7684\u667a\u80fd\u4f53\u80fd\u529b\u8d8a\u7a33\u5b9a\u53ef\u9760\u3002<\/li>\n<\/ol>\n","protected":false},"excerpt":{"rendered":"<p>MCPMark\u662f\u4e00\u4e2a\u7528\u4e8e\u8bc4\u4f30\u5927\u6a21\u578b\u667a\u80fd\u4f53\uff08Agentic\uff09\u80fd\u529b\u7684\u57fa\u51c6\u6d4b\u8bd5\u3002\u5b83\u901a\u8fc7\u5728\u4e00\u7cfb\u5217\u96c6\u6210\u4e86\u6a21\u578b\u4e0a\u4e0b\u6587\u534f\u8bae\uff08MCP\uff09\u7684\u771f\u5b9e\u8f6f\u4ef6\u73af\u5883\u4e2d\uff0c\u5bf9\u6a21\u578b\u8fdb\u884c\u538b\u529b\u6d4b\u8bd5\uff0c\u6765\u8861\u91cf\u5176\u81ea\u4e3b\u89c4\u5212\u3001\u63a8\u7406\u548c\u6267\u884c\u590d\u6742\u4efb\u52a1\u7684\u6c34\u5e73\u3002\u6d4b\u8bd5\u73af\u5883\u8986\u76d6\u4e86Notion\u3001GitHub&#8230;<\/p>\n","protected":false},"author":1,"featured_media":51095,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[425,496,20],"tags":[230],"class_list":["post-51088","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai-professional","category-ai-basic-services","category-tool","tag-aikaiyuanxiangmu"],"_links":{"self":[{"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/posts\/51088","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/comments?post=51088"}],"version-history":[{"count":0,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/posts\/51088\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/media\/51095"}],"wp:attachment":[{"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/media?parent=51088"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/categories?post=51088"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/tags?post=51088"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}