{"id":8247,"date":"2026-02-27T05:54:00","date_gmt":"2026-02-26T20:54:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=8247"},"modified":"2026-02-22T11:57:08","modified_gmt":"2026-02-22T02:57:08","slug":"what-makes-a-good-llm-agent-for-real-world-penetration-testing","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=8247","title":{"rendered":"What Makes a Good LLM Agent for Real-world Penetration Testing?\u00a0"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>What Makes a Good LLM Agent for Real-world Penetration Testing?\u00a0<\/strong>[37.6]<br>LLM\u3092\u30d9\u30fc\u30b9\u3068\u3057\u305f28\u306e\u6d78\u900f\u8a66\u9a13\u30b7\u30b9\u30c6\u30e0\u3092\u5206\u6790\u3057,\u8907\u96d1\u6027\u306e\u5897\u5927\u3092\u793a\u30593\u3064\u306e\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u30675\u3064\u306e\u4ee3\u8868\u7684\u5b9f\u88c5\u3092\u8a55\u4fa1\u3057\u305f\u3002 \u6211\u3005\u306f\u3001B\u578b\u969c\u5bb3\u304cLLM\u306e\u6839\u672c\u539f\u56e0\u3068\u307b\u3068\u3093\u3069\u5909\u308f\u3089\u305a\u3001\u30a8\u30fc\u30b8\u30a7\u30f3\u30c8\u306f\u30ea\u30a2\u30eb\u30bf\u30a4\u30e0\u306a\u30bf\u30b9\u30af\u306e\u96e3\u6613\u5ea6\u63a8\u5b9a\u3092\u6b20\u3044\u3066\u3044\u308b\u3053\u3068\u3092\u793a\u3059\u3002 Excalibur\u306f\u3001\u5f37\u529b\u306a\u30c4\u30fc\u30eb\u3068\u56f0\u96e3\u306a\u8a08\u753b\u3068\u3092\u7d50\u5408\u3057\u305f\u6d78\u900f\u8a66\u9a13\u30a8\u30fc\u30b8\u30a7\u30f3\u30c8\u3067\u3042\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2602.17622v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2602.17622v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Thu, 19 Feb 2026 18:42:40 GMT)<\/li>\n\n\n\n<li>\u30da\u30cd\u30c8\u30ec\u30fc\u30b7\u30e7\u30f3\u30c6\u30b9\u30c8\u3078\u306eLLMAgent\u9069\u7528\u3002<\/li>\n\n\n\n<li>\u300cPENTEST- GPT V2 achieves 91% task completion on CTF benchmarks (49% improvement over baselines) and compromises 4 of 5 hosts on the GOAD Active Directory environment versus 2 for prior systems\u300d\u3068\u3044\u3046\u7d50\u679c\u3002\u3053\u306e\u9818\u57df\u3082AI\u3068\u306e\u9023\u643a\u304c\u5fc5\u9808\u306b\u306a\u3063\u3066\u3044\u3066\u7d0d\u5f97\u611f\u304c\u3042\u308b\uff08\u3068\u540c\u6642\u306b\u6016\u3044\u3068\u3082\u611f\u3058\u308b\uff09<\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[352],"class_list":["post-8247","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-security"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/8247","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=8247"}],"version-history":[{"count":1,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/8247\/revisions"}],"predecessor-version":[{"id":8248,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/8247\/revisions\/8248"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=8247"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=8247"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=8247"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}