
{"id":11060,"date":"2026-08-14T01:39:12","date_gmt":"2026-08-13T17:39:12","guid":{"rendered":"https:\/\/infernews.com\/blog\/openart\/"},"modified":"2026-08-14T01:39:12","modified_gmt":"2026-08-13T17:39:12","slug":"openart","status":"publish","type":"post","link":"https:\/\/infernews.com\/blog\/openart\/","title":{"rendered":"OpenART\uff1a\u628a Agent \u5b89\u5168\u6e2c\u8a66\u642c\u9032\u53ef\u6f14\u5316\u74b0\u5883"},"content":{"rendered":"\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/infernews.com\/blog\/wp-content\/uploads\/2026\/08\/pasted-d0d4009f2e27.jpg\" alt=\"Repository image for AI45Lab\/OpenART\"><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">OpenART \u628a\u5de5\u5177\u578b code agent \u7684\u5b89\u5168\u6e2c\u8a66\uff0c\u62c9\u5230\u4e00\u500b\u66f4\u8cbc\u8fd1\u9577\u6d41\u7a0b\u5de5\u4f5c\u7684\u74b0\u5883\u88e1\u3002\u5b83\u4e0d\u662f\u55ae\u7d14\u9a57\u8b49\u6a21\u578b\u6703\u4e0d\u6703\u7b54\u5c0d\uff0c\u800c\u662f\u770b agent \u5728\u5171\u4eab\u72c0\u614b\u3001\u9023\u7e8c\u5de5\u5177\u64cd\u4f5c\u8207\u74b0\u5883\u6f14\u5316\u4e4b\u4e0b\uff0c\u6703\u4e0d\u6703\u6162\u6162\u8d70\u504f\uff0c\u9019\u5c0d\u505a\u5b89\u5168\u7814\u7a76\u3001\u7d05\u968a\u6e2c\u8a66\u6216 agent \u5e73\u53f0\u9a57\u8b49\u7684\u4eba\u90fd\u66f4\u6709\u53c3\u8003\u50f9\u503c\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u6574\u500b\u9805\u76ee\u662f Docker-native \u7684\u6846\u67b6\uff0c\u5b89\u88dd\u5f8c\u53ef\u5148\u8dd1\u672c\u5730 smoke task\uff0c\u78ba\u8a8d\u5bb9\u5668\u3001\u8a55\u4f30\u5668\u548c\u8def\u5f91\u90fd\u6b63\u5e38\uff0c\u518d\u5207\u63db\u5230\u9ad8\u8907\u96dc\u5ea6\u4efb\u52d9\u3002README \u5167\u4e5f\u63d0\u4f9b target-only \u8207 attacker \u5169\u7a2e\u8dd1\u6cd5\uff0c\u65b9\u4fbf\u5c0d\u6bd4\u55ae\u770b\u76ee\u6a19 agent \u8207\u52a0\u5165 OpenCode-compatible attacker \u7684\u5dee\u7570\uff1b\u4efb\u52d9\u5716\u6703\u5f9e openart-tools \u81ea\u52d5\u9078\u53d6\u6240\u9700\u5de5\u5177\uff0c\u4f7f\u7528\u65b9\u5f0f\u504f\u5411\u53ef\u63a7\u5be6\u9a57\u800c\u4e0d\u662f\u5373\u63d2\u5373\u7528\u670d\u52d9\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u5b83\u7684\u53d6\u6368\u5f88\u6e05\u695a\uff1aOpenART \u628a\u91cd\u9ede\u653e\u5728\u53ef\u91cd\u73fe\u3001\u53ef\u64f4\u5c55\u7684\u5b89\u5168\u8a55\u4f30\uff0c\u800c\u4e0d\u662f\u5305\u88dd\u6210\u4e00\u500b\u9762\u5411\u7d42\u7aef\u7528\u6236\u7684\u7522\u54c1\u3002\u9805\u76ee\u4fdd\u7559\u53ef\u57f7\u884c runtime\u3001\u5c11\u91cf\u9ad8\u8907\u96dc\u5ea6\u793a\u4f8b\u8207\u7ba1\u7406\u904e\u7684\u5de5\u5177\u5b50\u96c6\uff0c\u537b\u523b\u610f\u4e0d\u653e\u6279\u6b21\u5be6\u9a57\u9a45\u52d5\u3001\u751f\u6210\u7d50\u679c\u548c\u79c1\u6709\u5834\u666f\u8a9e\u6599\uff0c\u4ee3\u8868\u5b83\u66f4\u50cf\u7814\u7a76\u8207\u6e2c\u8a66\u5e95\u5ea7\uff0c\u9069\u5408\u9700\u8981\u81ea\u884c\u7de8\u6392\u5be6\u9a57\u7684\u4eba\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u8ad6\u6587\u80cc\u666f\u63d0\u5230\uff0cOpenART \u63d0\u4f9b\u8d85\u904e 10,000 \u500b\u5df2\u9a57\u8b49 stateful scenarios\uff0c\u6db5\u84cb 50 \u500b\u9818\u57df\uff0c\u4e26\u652f\u63f4 75 \u7a2e agent-model \u914d\u7f6e\u7684\u7d71\u4e00\u8a55\u4f30\uff1bEMHA\uff08Evolutionary Markov Hypergraph Attack\uff09\u5247\u662f\u7528\u4f86\u63a8\u52d5\u74b0\u5883\u6f14\u5316\u7684\u9ed1\u7bb1\u7b56\u7565\uff0c\u7121\u9700\u66f4\u65b0\u53c3\u6578\u3002\u6587\u4e2d\u4ea6\u6307\u51fa\uff0c\u8907\u96dc\u74b0\u5883\u4e0b\uff0c\u74b0\u5883\u6f14\u5316\u6bd4\u53ea\u6539\u6307\u4ee4\u66f4\u5bb9\u6613\u63ed\u793a\u5b89\u5168\u5931\u8aa4\uff0c\u9019\u6b63\u662f\u5b83\u76f8\u5c0d\u4e00\u822c\u77ed\u4efb\u52d9 benchmark \u7684\u95dc\u9375\u5dee\u5225\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u652f\u63f4\u5148\u505a\u672c\u5730 smoke run\uff0c\u518d\u8f49\u5165\u9ad8\u8907\u96dc\u5ea6\u4efb\u52d9\u9a57\u8b49<\/li>\n<li>\u4ee5 Docker \u8207\u4efb\u52d9\u5716\u7ba1\u7406\u5de5\u5177\uff0c\u4fbf\u65bc\u91cd\u73fe\u8207\u9694\u96e2\u74b0\u5883<\/li>\n<li>\u53ef\u6bd4\u8f03 target-only\u3001attacker \u8207\u4e0d\u540c agent-model \u914d\u7f6e<\/li>\n<li>\u9069\u5408\u505a agent \u5b89\u5168\u3001\u7d05\u968a\u6e2c\u8a66\u8207\u5e73\u53f0\u517c\u5bb9\u6027\u6aa2\u67e5<\/li>\n<li>\u5f37\u8abf\u9577\u6d41\u7a0b\u8207 stateful \u74b0\u5883\uff0c\u8f03\u80fd\u66b4\u9732\u7d2f\u7a4d\u6027\u98a8\u96aa<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/github.com\/AI45Lab\/OpenART\" rel=\"noopener noreferrer\"><strong>GitHub<\/strong><\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>OpenART \u5c07\u5de5\u5177\u578b code agent \u7684\u5b89\u5168\u8a55\u4f30\u505a\u6210\u53ef\u91cd\u73fe\u7684 Docker \u5de5\u4f5c\u6d41\uff0c\u91cd\u9ede\u4e0d\u662f\u8dd1\u55ae\u4e00\u4efb\u52d9\uff0c\u800c\u662f\u89c0\u5bdf\u74b0\u5883\u8b8a\u5316\u5982\u4f55\u7d2f\u7a4d\u6210\u98a8\u96aa\u3002\u4f60\u53ef\u4ee5\u628a\u5b83\u7406\u89e3\u6210\u4e00\u500b\u7528\u4f86\u505a agent red teaming \u7684\u958b\u6e90\u6846\u67b6\u3002<\/p>\n","protected":false},"author":8,"featured_media":11059,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"ai_generated_summary":"","footnotes":""},"categories":[133,116,170,199,200],"tags":[],"class_list":["post-11060","post","type-post","status-publish","format-standard","hentry","category-133","category-agentic","category-170","category-dataset-","category-200"],"_links":{"self":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/11060","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/users\/8"}],"replies":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/comments?post=11060"}],"version-history":[{"count":0,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/11060\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media\/11059"}],"wp:attachment":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media?parent=11060"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/categories?post=11060"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/tags?post=11060"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}