{"id":90741,"date":"2026-09-01T12:57:36","date_gmt":"2026-09-01T09:57:36","guid":{"rendered":"https:\/\/twodots.gr\/?p=90741"},"modified":"2026-09-01T12:57:36","modified_gmt":"2026-09-01T09:57:36","slug":"dreambench-swe-mnimi-ai-agents-mnimoniki-ygieini","status":"publish","type":"post","link":"https:\/\/twodots.gr\/en\/dreambench-swe-mnimi-ai-agents-mnimoniki-ygieini\/","title":{"rendered":"DreamBench-SWE: \u03b3\u03b9\u03b1\u03c4\u03af \u03b7 \u03bc\u03bd\u03ae\u03bc\u03b7 \u03c4\u03c9\u03bd AI agents \u03c7\u03c1\u03b5\u03b9\u03ac\u03b6\u03b5\u03c4\u03b1\u03b9 \u03c5\u03b3\u03b9\u03b5\u03b9\u03bd\u03ae, \u03cc\u03c7\u03b9 \u03b1\u03c0\u03bb\u03ce\u03c2 \u03c7\u03c9\u03c1\u03b7\u03c4\u03b9\u03ba\u03cc\u03c4\u03b7\u03c4\u03b1"},"content":{"rendered":"<div class=\"td-article-lede\">\n<p><strong>\u03a4\u03bf DreamBench-SWE \u03b4\u03b5\u03af\u03c7\u03bd\u03b5\u03b9 \u03cc\u03c4\u03b9 \u03b7 \u03c0\u03b5\u03c1\u03b9\u03c3\u03c3\u03cc\u03c4\u03b5\u03c1\u03b7 \u03bc\u03bd\u03ae\u03bc\u03b7 \u03b4\u03b5\u03bd \u03ba\u03ac\u03bd\u03b5\u03b9 \u03b1\u03c5\u03c4\u03cc\u03bc\u03b1\u03c4\u03b1 \u03ad\u03bd\u03b1\u03bd AI agent \u03c0\u03b9\u03bf \u03b1\u03be\u03b9\u03cc\u03c0\u03b9\u03c3\u03c4\u03bf.<\/strong> \u03a3\u03b5 \u03b5\u03c1\u03b3\u03b1\u03c3\u03af\u03b5\u03c2 \u03ba\u03ce\u03b4\u03b9\u03ba\u03b1 \u03c0\u03bf\u03c5 \u03c3\u03c5\u03bd\u03b5\u03c7\u03af\u03b6\u03bf\u03bd\u03c4\u03b1\u03b9 \u03c3\u03b5 \u03c0\u03bf\u03bb\u03bb\u03ad\u03c2 \u03c3\u03c5\u03bd\u03b5\u03b4\u03c1\u03af\u03b5\u03c2, \u03b7 \u03b1\u03be\u03af\u03b1 \u03b2\u03c1\u03af\u03c3\u03ba\u03b5\u03c4\u03b1\u03b9 \u03c3\u03c4\u03b7 \u03bc\u03bd\u03b7\u03bc\u03bf\u03bd\u03b9\u03ba\u03ae \u03c5\u03b3\u03b9\u03b5\u03b9\u03bd\u03ae: \u03c3\u03c9\u03c3\u03c4\u03ae \u03c0\u03c1\u03bf\u03ad\u03bb\u03b5\u03c5\u03c3\u03b7, \u03c0\u03b5\u03c1\u03b9\u03bf\u03c1\u03b9\u03c3\u03bc\u03ad\u03bd\u03bf scope, \u03b1\u03bd\u03b1\u03b3\u03bd\u03ce\u03c1\u03b9\u03c3\u03b7 \u03c0\u03b1\u03c1\u03c9\u03c7\u03b7\u03bc\u03ad\u03bd\u03c9\u03bd \u03c3\u03c4\u03bf\u03b9\u03c7\u03b5\u03af\u03c9\u03bd, \u03b1\u03c3\u03c6\u03b1\u03bb\u03ae\u03c2 \u03b1\u03bd\u03ac\u03ba\u03c4\u03b7\u03c3\u03b7 \u03ba\u03b1\u03b9 \u03b5\u03ba\u03c4\u03b5\u03bb\u03ad\u03c3\u03b9\u03bc\u03b7 \u03b5\u03c0\u03b1\u03bb\u03ae\u03b8\u03b5\u03c5\u03c3\u03b7 \u03c4\u03b7\u03c2 \u03c4\u03b5\u03bb\u03b9\u03ba\u03ae\u03c2 \u03b1\u03bb\u03bb\u03b1\u03b3\u03ae\u03c2.<\/p>\n<\/div>\n<div class=\"td-article-toc\">\n<div class=\"td-toc-title\">Contents<\/div>\n<ul>\n<li><a href=\"#mnimi-polles-synedries\">\u0393\u03b9\u03b1\u03c4\u03af \u03b7 \u03bc\u03bd\u03ae\u03bc\u03b7 \u03b3\u03af\u03bd\u03b5\u03c4\u03b1\u03b9 \u03c1\u03af\u03c3\u03ba\u03bf \u03c3\u03b5 \u03c0\u03bf\u03bb\u03bb\u03ad\u03c2 \u03c3\u03c5\u03bd\u03b5\u03b4\u03c1\u03af\u03b5\u03c2<\/a><\/li>\n<li><a href=\"#ti-metra-dreambench-swe\">\u03a4\u03b9 \u03bc\u03b5\u03c4\u03c1\u03ac \u03b4\u03b9\u03b1\u03c6\u03bf\u03c1\u03b5\u03c4\u03b9\u03ba\u03ac \u03c4\u03bf DreamBench-SWE<\/a><\/li>\n<li><a href=\"#pagides-mnimonikis-ygieinis\">\u039f\u03b9 \u03c0\u03b1\u03b3\u03af\u03b4\u03b5\u03c2 \u03c0\u03bf\u03c5 \u03b1\u03c0\u03bf\u03ba\u03b1\u03bb\u03cd\u03c0\u03c4\u03bf\u03c5\u03bd \u03ba\u03b1\u03ba\u03ae \u03bc\u03bd\u03b7\u03bc\u03bf\u03bd\u03b9\u03ba\u03ae \u03c5\u03b3\u03b9\u03b5\u03b9\u03bd\u03ae<\/a><\/li>\n<li><a href=\"#raw-typed-retrieval-gate\">Raw evidence, typed memories \u03ba\u03b1\u03b9 retrieval gate<\/a><\/li>\n<li><a href=\"#apotelesmata-v2-v21\">\u03a4\u03b9 \u03ad\u03b4\u03b5\u03b9\u03be\u03b1\u03bd \u03c4\u03b1 v2 \u03ba\u03b1\u03b9 v2.1<\/a><\/li>\n<li><a href=\"#peiramatikoi-elegchoi\">\u03a0\u03ce\u03c2 \u03bf\u03b9 \u03c0\u03b5\u03b9\u03c1\u03b1\u03bc\u03b1\u03c4\u03b9\u03ba\u03bf\u03af \u03ad\u03bb\u03b5\u03b3\u03c7\u03bf\u03b9 \u03ba\u03c1\u03b1\u03c4\u03bf\u03cd\u03bd \u03c4\u03bf benchmark \u03c4\u03af\u03bc\u03b9\u03bf<\/a><\/li>\n<li><a href=\"#diakyvernisi-mnimis-paragogi\">\u0388\u03bd\u03b1 \u03bc\u03bf\u03bd\u03c4\u03ad\u03bb\u03bf \u03b4\u03b9\u03b1\u03ba\u03c5\u03b2\u03ad\u03c1\u03bd\u03b7\u03c3\u03b7\u03c2 \u03bc\u03bd\u03ae\u03bc\u03b7\u03c2 \u03b3\u03b9\u03b1 \u03c0\u03b1\u03c1\u03b1\u03b3\u03c9\u03b3\u03ae<\/a><\/li>\n<li><a href=\"#pilot-axiologisis-mnimis\">\u03a0\u03ce\u03c2 \u03c3\u03c4\u03ae\u03bd\u03b5\u03c4\u03b1\u03b9 pilot \u03b1\u03be\u03b9\u03bf\u03bb\u03cc\u03b3\u03b7\u03c3\u03b7\u03c2 \u03bc\u03bd\u03ae\u03bc\u03b7\u03c2<\/a><\/li>\n<li><a href=\"#kostos-asfaleia\">\u039a\u03cc\u03c3\u03c4\u03bf\u03c2, \u03b1\u03c3\u03c6\u03ac\u03bb\u03b5\u03b9\u03b1 \u03ba\u03b1\u03b9 \u03b5\u03c0\u03b9\u03c7\u03b5\u03b9\u03c1\u03b7\u03c3\u03b9\u03b1\u03ba\u03cc \u03b1\u03c0\u03bf\u03c4\u03ad\u03bb\u03b5\u03c3\u03bc\u03b1<\/a><\/li>\n<li><a href=\"#oria-kai-symperasma\">\u03a4\u03b1 \u03cc\u03c1\u03b9\u03b1 \u03ba\u03b1\u03b9 \u03c4\u03bf \u03c3\u03c4\u03c1\u03b1\u03c4\u03b7\u03b3\u03b9\u03ba\u03cc \u03c3\u03c5\u03bc\u03c0\u03ad\u03c1\u03b1\u03c3\u03bc\u03b1<\/a><\/li>\n<\/ul>\n<\/div>\n<p>\u0388\u03bd\u03b1\u03c2 AI agent \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b8\u03c5\u03bc\u03ac\u03c4\u03b1\u03b9 \u03c0\u03bf\u03bb\u03bb\u03ac \u03ba\u03b1\u03b9 \u03c0\u03b1\u03c1\u2019 \u03cc\u03bb\u03b1 \u03b1\u03c5\u03c4\u03ac \u03bd\u03b1 \u03ba\u03ac\u03bd\u03b5\u03b9 \u03c7\u03b5\u03b9\u03c1\u03cc\u03c4\u03b5\u03c1\u03b7 \u03b4\u03bf\u03c5\u03bb\u03b5\u03b9\u03ac. \u039c\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b1\u03bd\u03b1\u03c3\u03cd\u03c1\u03b5\u03b9 \u03bc\u03b9\u03b1 \u03c0\u03b1\u03bb\u03b9\u03ac \u03b1\u03c1\u03c7\u03b9\u03c4\u03b5\u03ba\u03c4\u03bf\u03bd\u03b9\u03ba\u03ae \u03b1\u03c0\u03cc\u03c6\u03b1\u03c3\u03b7 \u03c0\u03bf\u03c5 \u03ad\u03c7\u03b5\u03b9 \u03b1\u03ba\u03c5\u03c1\u03c9\u03b8\u03b5\u03af, \u03bd\u03b1 \u03b5\u03c6\u03b1\u03c1\u03bc\u03cc\u03c3\u03b5\u03b9 \u03c0\u03b1\u03c1\u03b1\u03c4\u03ae\u03c1\u03b7\u03c3\u03b7 reviewer \u03c3\u03b5 \u03bb\u03ac\u03b8\u03bf\u03c2 module \u03ae \u03bd\u03b1 \u03bc\u03b5\u03c4\u03b1\u03c4\u03c1\u03ad\u03c8\u03b5\u03b9 \u03bc\u03b9\u03b1 \u03b1\u03b2\u03ad\u03b2\u03b1\u03b9\u03b7 \u03b4\u03b9\u03ac\u03b3\u03bd\u03c9\u03c3\u03b7 \u03c3\u03b5 \u03bc\u03cc\u03bd\u03b9\u03bc\u03bf \u03ba\u03b1\u03bd\u03cc\u03bd\u03b1. \u03a4\u03bf \u03c0\u03c1\u03cc\u03b2\u03bb\u03b7\u03bc\u03b1 \u03b4\u03b5\u03bd \u03b5\u03af\u03bd\u03b1\u03b9 \u03bc\u03cc\u03bd\u03bf \u03c0\u03cc\u03c3\u03b7 \u03bc\u03bd\u03ae\u03bc\u03b7 \u03b4\u03b9\u03b1\u03b8\u03ad\u03c4\u03b5\u03b9 \u03c4\u03bf \u03c3\u03cd\u03c3\u03c4\u03b7\u03bc\u03b1, \u03b1\u03bb\u03bb\u03ac \u03b1\u03bd \u03b8\u03c5\u03bc\u03ac\u03c4\u03b1\u03b9 \u03c4\u03bf \u03c3\u03c9\u03c3\u03c4\u03cc \u03c3\u03c4\u03bf\u03b9\u03c7\u03b5\u03af\u03bf, \u03bc\u03b5 \u03c4\u03b7 \u03c3\u03c9\u03c3\u03c4\u03ae \u03c0\u03c1\u03bf\u03ad\u03bb\u03b5\u03c5\u03c3\u03b7, \u03c3\u03c4\u03bf \u03c3\u03c9\u03c3\u03c4\u03cc \u03c0\u03bb\u03b1\u03af\u03c3\u03b9\u03bf \u03ba\u03b1\u03b9 \u03c4\u03b7 \u03c3\u03c9\u03c3\u03c4\u03ae \u03c3\u03c4\u03b9\u03b3\u03bc\u03ae.<\/p>\n<p>\u0391\u03c5\u03c4\u03cc \u03b5\u03c0\u03b9\u03c7\u03b5\u03b9\u03c1\u03b5\u03af \u03bd\u03b1 \u03bc\u03b5\u03c4\u03c1\u03ae\u03c3\u03b5\u03b9 \u03c4\u03bf <strong>DreamBench-SWE<\/strong>, \u03ad\u03bd\u03b1 benchmark \u03b3\u03b9\u03b1 \u03c4\u03b7 \u03bc\u03bd\u03b7\u03bc\u03bf\u03bd\u03b9\u03ba\u03ae \u03c5\u03b3\u03b9\u03b5\u03b9\u03bd\u03ae software agents \u03c0\u03bf\u03c5 \u03b5\u03c1\u03b3\u03ac\u03b6\u03bf\u03bd\u03c4\u03b1\u03b9 \u03c3\u03b5 \u03b4\u03b9\u03b1\u03b4\u03bf\u03c7\u03b9\u03ba\u03ad\u03c2 \u03c3\u03c5\u03bd\u03b5\u03b4\u03c1\u03af\u03b5\u03c2. \u0394\u03b5\u03bd \u03b5\u03af\u03bd\u03b1\u03b9 \u03b3\u03b5\u03bd\u03b9\u03ba\u03ae \u03ba\u03b1\u03c4\u03ac\u03c4\u03b1\u03be\u03b7 \u03c0\u03c1\u03bf\u03ca\u03cc\u03bd\u03c4\u03c9\u03bd \u03bc\u03bd\u03ae\u03bc\u03b7\u03c2 \u03ba\u03b1\u03b9 \u03b4\u03b5\u03bd \u03b1\u03c0\u03bf\u03b4\u03b5\u03b9\u03ba\u03bd\u03cd\u03b5\u03b9 \u03cc\u03c4\u03b9 \u03bc\u03af\u03b1 \u03b1\u03c1\u03c7\u03b9\u03c4\u03b5\u03ba\u03c4\u03bf\u03bd\u03b9\u03ba\u03ae \u03b5\u03af\u03bd\u03b1\u03b9 \u03ba\u03b1\u03b8\u03bf\u03bb\u03b9\u03ba\u03ac \u03b1\u03bd\u03ce\u03c4\u03b5\u03c1\u03b7. \u03a0\u03c1\u03bf\u03c4\u03b5\u03af\u03bd\u03b5\u03b9 \u03cc\u03bc\u03c9\u03c2 \u03b1\u03c5\u03c3\u03c4\u03b7\u03c1\u03cc \u03c4\u03c1\u03cc\u03c0\u03bf \u03bd\u03b1 \u03b5\u03bb\u03b5\u03b3\u03c7\u03b8\u03b5\u03af \u03b1\u03bd \u03b7 \u03bc\u03bd\u03ae\u03bc\u03b7 \u03bf\u03b4\u03b7\u03b3\u03b5\u03af \u03c3\u03b5 \u03c0\u03c1\u03b1\u03b3\u03bc\u03b1\u03c4\u03b9\u03ba\u03ac \u03c3\u03c9\u03c3\u03c4\u03ae \u03b5\u03c1\u03b3\u03b1\u03c3\u03af\u03b1 \u03c0\u03ac\u03bd\u03c9 \u03c3\u03b5 \u03ba\u03ce\u03b4\u03b9\u03ba\u03b1. \u0397 \u03b4\u03b9\u03ac\u03ba\u03c1\u03b9\u03c3\u03b7 \u03ad\u03c7\u03b5\u03b9 \u03c0\u03c1\u03b1\u03ba\u03c4\u03b9\u03ba\u03ae \u03b1\u03be\u03af\u03b1 \u03b3\u03b9\u03b1 \u03c4\u03b9\u03c2 <a href=\"https:\/\/twodots.gr\/ai-agents-paragogi-leitourgikes-astochies\/\">\u03bb\u03b5\u03b9\u03c4\u03bf\u03c5\u03c1\u03b3\u03b9\u03ba\u03ad\u03c2 \u03b1\u03c3\u03c4\u03bf\u03c7\u03af\u03b5\u03c2 AI agents \u03c3\u03c4\u03b7\u03bd \u03c0\u03b1\u03c1\u03b1\u03b3\u03c9\u03b3\u03ae<\/a>, \u03cc\u03c0\u03bf\u03c5 \u03ad\u03bd\u03b1 \u03c0\u03b5\u03b9\u03c3\u03c4\u03b9\u03ba\u03cc \u03b1\u03c0\u03bf\u03c4\u03ad\u03bb\u03b5\u03c3\u03bc\u03b1 \u03b4\u03b5\u03bd \u03b1\u03c1\u03ba\u03b5\u03af \u03c7\u03c9\u03c1\u03af\u03c2 \u03ad\u03bb\u03b5\u03b3\u03c7\u03bf \u03c4\u03b7\u03c2 \u03b4\u03b9\u03b1\u03b4\u03c1\u03bf\u03bc\u03ae\u03c2.<\/p>\n<h2 id=\"mnimi-polles-synedries\">\u0393\u03b9\u03b1\u03c4\u03af \u03b7 \u03bc\u03bd\u03ae\u03bc\u03b7 \u03b3\u03af\u03bd\u03b5\u03c4\u03b1\u03b9 \u03c1\u03af\u03c3\u03ba\u03bf \u03c3\u03b5 \u03c0\u03bf\u03bb\u03bb\u03ad\u03c2 \u03c3\u03c5\u03bd\u03b5\u03b4\u03c1\u03af\u03b5\u03c2<\/h2>\n<p>\u039f\u03b9 \u03c3\u03cd\u03b3\u03c7\u03c1\u03bf\u03bd\u03bf\u03b9 software agents \u03b4\u03b5\u03bd \u03c0\u03b5\u03c1\u03b9\u03bf\u03c1\u03af\u03b6\u03bf\u03bd\u03c4\u03b1\u03b9 \u03c0\u03ac\u03bd\u03c4\u03b1 \u03c3\u03b5 \u03ad\u03bd\u03b1 prompt \u03ba\u03b1\u03b9 \u03bc\u03af\u03b1 \u03b1\u03c0\u03ac\u03bd\u03c4\u03b7\u03c3\u03b7. \u0395\u03c0\u03b9\u03c3\u03c4\u03c1\u03ad\u03c6\u03bf\u03c5\u03bd \u03c3\u03c4\u03bf \u03af\u03b4\u03b9\u03bf repository, \u03b4\u03b9\u03b1\u03b2\u03ac\u03b6\u03bf\u03c5\u03bd \u03c0\u03c1\u03bf\u03b7\u03b3\u03bf\u03cd\u03bc\u03b5\u03bd\u03b5\u03c2 \u03b1\u03bb\u03bb\u03b1\u03b3\u03ad\u03c2, \u03bb\u03b1\u03bc\u03b2\u03ac\u03bd\u03bf\u03c5\u03bd feedback \u03ba\u03b1\u03b9 \u03c3\u03c5\u03bd\u03b5\u03c7\u03af\u03b6\u03bf\u03c5\u03bd \u03b5\u03c1\u03b3\u03b1\u03c3\u03af\u03b5\u03c2 \u03c0\u03bf\u03c5 \u03be\u03b5\u03ba\u03af\u03bd\u03b7\u03c3\u03b1\u03bd \u03ce\u03c1\u03b5\u03c2 \u03ae \u03b7\u03bc\u03ad\u03c1\u03b5\u03c2 \u03c0\u03c1\u03b9\u03bd. \u03a3\u03b5 \u03b1\u03c5\u03c4\u03cc \u03c4\u03bf \u03c0\u03b5\u03c1\u03b9\u03b2\u03ac\u03bb\u03bb\u03bf\u03bd, \u03c0\u03bb\u03b7\u03c1\u03bf\u03c6\u03bf\u03c1\u03af\u03b5\u03c2 \u03c0\u03bf\u03c5 \u03b1\u03c0\u03bf\u03c5\u03c3\u03b9\u03ac\u03b6\u03bf\u03c5\u03bd \u03b1\u03c0\u03cc \u03c4\u03bf \u03c4\u03c1\u03ad\u03c7\u03bf\u03bd context \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b5\u03af\u03bd\u03b1\u03b9 \u03b1\u03bd\u03b1\u03b3\u03ba\u03b1\u03af\u03b5\u03c2: \u03ad\u03bd\u03b1 generated file \u03b4\u03b5\u03bd \u03c0\u03c1\u03ad\u03c0\u03b5\u03b9 \u03bd\u03b1 \u03c4\u03c1\u03bf\u03c0\u03bf\u03c0\u03bf\u03b9\u03b7\u03b8\u03b5\u03af \u03b1\u03c0\u03b5\u03c5\u03b8\u03b5\u03af\u03b1\u03c2, \u03ad\u03bd\u03b1\u03c2 \u03ba\u03b1\u03bd\u03cc\u03bd\u03b1\u03c2 \u03b1\u03c6\u03bf\u03c1\u03ac \u03bc\u03cc\u03bd\u03bf \u03ad\u03bd\u03b1 module \u03ae \u03bc\u03b9\u03b1 \u03c0\u03b1\u03bb\u03b9\u03ac \u03c5\u03c0\u03cc\u03b8\u03b5\u03c3\u03b7 \u03ad\u03c7\u03b5\u03b9 \u03b1\u03bd\u03c4\u03b9\u03ba\u03b1\u03c4\u03b1\u03c3\u03c4\u03b1\u03b8\u03b5\u03af \u03b1\u03c0\u03cc \u03bd\u03b5\u03cc\u03c4\u03b5\u03c1\u03b7 \u03b1\u03c0\u03cc\u03c6\u03b1\u03c3\u03b7.<\/p>\n<p>\u0397 \u03b1\u03c0\u03bb\u03ae \u03b1\u03c0\u03bf\u03b8\u03ae\u03ba\u03b5\u03c5\u03c3\u03b7 \u03cc\u03bb\u03c9\u03bd \u03c4\u03c9\u03bd \u03c0\u03c1\u03bf\u03b7\u03b3\u03bf\u03cd\u03bc\u03b5\u03bd\u03c9\u03bd \u03b4\u03b5\u03b4\u03bf\u03bc\u03ad\u03bd\u03c9\u03bd \u03b4\u03b5\u03bd \u03bb\u03cd\u03bd\u03b5\u03b9 \u03c4\u03bf \u03c0\u03c1\u03cc\u03b2\u03bb\u03b7\u03bc\u03b1. \u0397 \u03b1\u03bd\u03ac\u03ba\u03c4\u03b7\u03c3\u03b7 \u03c0\u03b1\u03c1\u03cc\u03bc\u03bf\u03b9\u03bf\u03c5 \u03b1\u03bb\u03bb\u03ac \u03c0\u03b1\u03c1\u03c9\u03c7\u03b7\u03bc\u03ad\u03bd\u03bf\u03c5 \u03c3\u03c4\u03bf\u03b9\u03c7\u03b5\u03af\u03bf\u03c5 \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03bf\u03b4\u03b7\u03b3\u03ae\u03c3\u03b5\u03b9 \u03c3\u03b5 \u03bb\u03ac\u03b8\u03bf\u03c2 patch. \u0397 \u03c5\u03c0\u03b5\u03c1\u03b3\u03b5\u03bd\u03af\u03ba\u03b5\u03c5\u03c3\u03b7 \u03b5\u03bd\u03cc\u03c2 \u03c3\u03c7\u03bf\u03bb\u03af\u03bf\u03c5 \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b5\u03c0\u03b7\u03c1\u03b5\u03ac\u03c3\u03b5\u03b9 \u03ac\u03c3\u03c7\u03b5\u03c4\u03b1 \u03c4\u03bc\u03ae\u03bc\u03b1\u03c4\u03b1 \u03c4\u03bf\u03c5 \u03ba\u03ce\u03b4\u03b9\u03ba\u03b1. \u039c\u03b9\u03b1 \u03b1\u03c0\u03bf\u03c4\u03c5\u03c7\u03b7\u03bc\u03ad\u03bd\u03b7 \u03b4\u03bf\u03ba\u03b9\u03bc\u03ae \u03c0\u03bf\u03c5 \u03bf\u03c6\u03b5\u03b9\u03bb\u03cc\u03c4\u03b1\u03bd \u03c3\u03b5 \u03b1\u03c3\u03c4\u03ac\u03b8\u03b5\u03b9\u03b1 \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b1\u03c0\u03bf\u03b8\u03b7\u03ba\u03b5\u03c5\u03c4\u03b5\u03af \u03bb\u03b1\u03bd\u03b8\u03b1\u03c3\u03bc\u03ad\u03bd\u03b1 \u03c9\u03c2 \u03b1\u03b9\u03c4\u03b9\u03ce\u03b4\u03b7\u03c2 \u03ba\u03b1\u03bd\u03cc\u03bd\u03b1\u03c2.<\/p>\n<aside class=\"td-article-note\">\n<p><strong>Memory hygiene \u03b4\u03b5\u03bd \u03c3\u03b7\u03bc\u03b1\u03af\u03bd\u03b5\u03b9 \u00ab\u03b8\u03c5\u03bc\u03ae\u03c3\u03bf\u03c5 \u03c0\u03b5\u03c1\u03b9\u03c3\u03c3\u03cc\u03c4\u03b5\u03c1\u03b1\u00bb.<\/strong> \u03a3\u03b7\u03bc\u03b1\u03af\u03bd\u03b5\u03b9 \u03cc\u03c4\u03b9 \u03c4\u03bf \u03c3\u03cd\u03c3\u03c4\u03b7\u03bc\u03b1 \u03b4\u03b9\u03b1\u03c4\u03b7\u03c1\u03b5\u03af \u03c4\u03b7\u03bd \u03c0\u03c1\u03c9\u03c4\u03bf\u03b3\u03b5\u03bd\u03ae \u03b1\u03c0\u03cc\u03b4\u03b5\u03b9\u03be\u03b7, \u03b1\u03bd\u03b1\u03b3\u03bd\u03c9\u03c1\u03af\u03b6\u03b5\u03b9 \u03c0\u03cc\u03c4\u03b5 \u03bc\u03b9\u03b1 \u03c0\u03bb\u03b7\u03c1\u03bf\u03c6\u03bf\u03c1\u03af\u03b1 \u03ad\u03c0\u03b1\u03c8\u03b5 \u03bd\u03b1 \u03b9\u03c3\u03c7\u03cd\u03b5\u03b9, \u03c0\u03b5\u03c1\u03b9\u03bf\u03c1\u03af\u03b6\u03b5\u03b9 \u03ba\u03ac\u03b8\u03b5 \u03ba\u03b1\u03bd\u03cc\u03bd\u03b1 \u03c3\u03c4\u03bf \u03c3\u03c9\u03c3\u03c4\u03cc scope \u03ba\u03b1\u03b9 \u03ba\u03c1\u03b1\u03c4\u03ac \u03ad\u03be\u03c9 \u03b1\u03c0\u03cc \u03c4\u03bf context \u03cc,\u03c4\u03b9 \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03c0\u03b1\u03c1\u03b1\u03c0\u03bb\u03b1\u03bd\u03ae\u03c3\u03b5\u03b9 \u03c4\u03b7\u03bd \u03c4\u03c1\u03ad\u03c7\u03bf\u03c5\u03c3\u03b1 \u03b5\u03c1\u03b3\u03b1\u03c3\u03af\u03b1.<\/p>\n<\/aside>\n<p>\u0391\u03c5\u03c4\u03ae \u03b7 \u03c0\u03b5\u03b9\u03b8\u03b1\u03c1\u03c7\u03af\u03b1 \u03b1\u03c6\u03bf\u03c1\u03ac \u03c4\u03bf \u03c0\u03bb\u03ae\u03c1\u03b5\u03c2 agent system, \u03cc\u03c7\u03b9 \u03bc\u03cc\u03bd\u03bf \u03c4\u03bf \u03bc\u03bf\u03bd\u03c4\u03ad\u03bb\u03bf. \u03a4\u03bf harness, \u03c4\u03bf event store, \u03c4\u03bf consolidation layer \u03ba\u03b1\u03b9 \u03b7 \u03c0\u03bf\u03bb\u03b9\u03c4\u03b9\u03ba\u03ae \u03b1\u03bd\u03ac\u03ba\u03c4\u03b7\u03c3\u03b7\u03c2 \u03ba\u03b1\u03b8\u03bf\u03c1\u03af\u03b6\u03bf\u03c5\u03bd \u03c4\u03b9 \u03b8\u03b1 \u03b5\u03bc\u03c6\u03b1\u03bd\u03b9\u03c3\u03c4\u03b5\u03af \u03c3\u03c4\u03b7\u03bd \u03b5\u03c0\u03cc\u03bc\u03b5\u03bd\u03b7 \u03c3\u03c5\u03bd\u03b5\u03b4\u03c1\u03af\u03b1. \u0393\u03b9\u2019 \u03b1\u03c5\u03c4\u03cc \u03b7 <a href=\"https:\/\/twodots.gr\/enterprise-ai-harnesses-diakyvernisi-architektoniki\/\">\u03b4\u03b9\u03b1\u03ba\u03c5\u03b2\u03ad\u03c1\u03bd\u03b7\u03c3\u03b7 \u03c4\u03c9\u03bd enterprise AI harnesses<\/a> \u03c0\u03c1\u03ad\u03c0\u03b5\u03b9 \u03bd\u03b1 \u03c3\u03c7\u03b5\u03b4\u03b9\u03ac\u03b6\u03b5\u03c4\u03b1\u03b9 \u03c9\u03c2 \u03bc\u03ad\u03c1\u03bf\u03c2 \u03c4\u03b7\u03c2 \u03b1\u03c1\u03c7\u03b9\u03c4\u03b5\u03ba\u03c4\u03bf\u03bd\u03b9\u03ba\u03ae\u03c2 \u03ba\u03b1\u03b9 \u03cc\u03c7\u03b9 \u03c9\u03c2 \u03bc\u03b5\u03c4\u03b1\u03b3\u03b5\u03bd\u03ad\u03c3\u03c4\u03b5\u03c1\u03bf checklist.<\/p>\n<h2 id=\"ti-metra-dreambench-swe\">\u03a4\u03b9 \u03bc\u03b5\u03c4\u03c1\u03ac \u03b4\u03b9\u03b1\u03c6\u03bf\u03c1\u03b5\u03c4\u03b9\u03ba\u03ac \u03c4\u03bf DreamBench-SWE<\/h2>\n<p>\u03a0\u03bf\u03bb\u03bb\u03ac benchmarks \u03b5\u03be\u03b5\u03c4\u03ac\u03b6\u03bf\u03c5\u03bd \u03b1\u03bd \u03ad\u03bd\u03b1 \u03bc\u03bf\u03bd\u03c4\u03ad\u03bb\u03bf \u03b1\u03bd\u03b1\u03ba\u03b1\u03bb\u03b5\u03af \u03c0\u03bb\u03b7\u03c1\u03bf\u03c6\u03bf\u03c1\u03af\u03b5\u03c2 \u03b1\u03c0\u03cc \u03bc\u03b5\u03b3\u03ac\u03bb\u03b7 \u03c3\u03c5\u03bd\u03bf\u03bc\u03b9\u03bb\u03af\u03b1 \u03ae \u03b1\u03bd \u03b1\u03c0\u03b1\u03bd\u03c4\u03ac \u03c3\u03c9\u03c3\u03c4\u03ac \u03c3\u03b5 \u03b5\u03c1\u03c9\u03c4\u03ae\u03c3\u03b5\u03b9\u03c2. \u0395\u03b4\u03ce \u03b7 \u03b1\u03c0\u03b1\u03af\u03c4\u03b7\u03c3\u03b7 \u03b5\u03af\u03bd\u03b1\u03b9 \u03b1\u03c5\u03c3\u03c4\u03b7\u03c1\u03cc\u03c4\u03b5\u03c1\u03b7. \u039a\u03ac\u03b8\u03b5 trap \u03b5\u03ba\u03c4\u03b5\u03af\u03bd\u03b5\u03c4\u03b1\u03b9 \u03c3\u03b5 \u03c3\u03c5\u03bd\u03b5\u03b4\u03c1\u03af\u03b5\u03c2 S1, S2 \u03ba\u03b1\u03b9 S3. \u0397 \u03ba\u03c1\u03af\u03c3\u03b9\u03bc\u03b7 \u03c0\u03bb\u03b7\u03c1\u03bf\u03c6\u03bf\u03c1\u03af\u03b1 \u03b5\u03bc\u03c6\u03b1\u03bd\u03af\u03b6\u03b5\u03c4\u03b1\u03b9 \u03bd\u03c9\u03c1\u03af\u03c4\u03b5\u03c1\u03b1, \u03b4\u03b5\u03bd \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03c3\u03c5\u03bd\u03b1\u03c7\u03b8\u03b5\u03af \u03b1\u03be\u03b9\u03cc\u03c0\u03b9\u03c3\u03c4\u03b1 \u03b1\u03c0\u03cc \u03c4\u03bf \u03c4\u03b5\u03bb\u03b5\u03c5\u03c4\u03b1\u03af\u03bf prompt \u03ba\u03b1\u03b9 \u03b4\u03b5\u03bd \u03c5\u03c0\u03ac\u03c1\u03c7\u03b5\u03b9 \u03c3\u03c4\u03bf filesystem \u03c4\u03bf\u03c5 wake agent \u03cc\u03c4\u03b1\u03bd \u03c6\u03c4\u03ac\u03bd\u03b5\u03b9 \u03b7 \u03c4\u03b5\u03bb\u03b9\u03ba\u03ae \u03b5\u03c1\u03b3\u03b1\u03c3\u03af\u03b1.<\/p>\n<p>\u0397 \u03c4\u03b5\u03bb\u03b9\u03ba\u03ae \u03b1\u03c0\u03ac\u03bd\u03c4\u03b7\u03c3\u03b7 \u03b4\u03b5\u03bd \u03b2\u03b1\u03b8\u03bc\u03bf\u03bb\u03bf\u03b3\u03b5\u03af\u03c4\u03b1\u03b9 \u03b1\u03c0\u03cc \u03b3\u03bb\u03c9\u03c3\u03c3\u03b9\u03ba\u03cc \u03ba\u03c1\u03b9\u03c4\u03ae \u03c0\u03bf\u03c5 \u03b1\u03c0\u03bf\u03c6\u03b1\u03c3\u03af\u03b6\u03b5\u03b9 \u03b1\u03bd \u00ab\u03b1\u03ba\u03bf\u03cd\u03b3\u03b5\u03c4\u03b1\u03b9 \u03c3\u03c9\u03c3\u03c4\u03ae\u00bb. \u039f agent \u03c0\u03c1\u03ad\u03c0\u03b5\u03b9 \u03bd\u03b1 \u03ba\u03ac\u03bd\u03b5\u03b9 \u03c0\u03c1\u03b1\u03b3\u03bc\u03b1\u03c4\u03b9\u03ba\u03ae \u03b1\u03bb\u03bb\u03b1\u03b3\u03ae \u03c3\u03b5 repository \u03ba\u03b1\u03b9 \u03b7 \u03b5\u03c0\u03b9\u03c4\u03c5\u03c7\u03af\u03b1 \u03b5\u03bb\u03ad\u03b3\u03c7\u03b5\u03c4\u03b1\u03b9 \u03b1\u03c0\u03cc \u03ba\u03c1\u03c5\u03c6\u03cc, \u03b5\u03ba\u03c4\u03b5\u03bb\u03ad\u03c3\u03b9\u03bc\u03bf oracle. \u0388\u03bd\u03b1 trap \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b1\u03c0\u03b1\u03b9\u03c4\u03b5\u03af \u03c4\u03b7\u03bd \u03ba\u03b1\u03c4\u03ac \u03b3\u03c1\u03ac\u03bc\u03bc\u03b1 \u03b5\u03c0\u03b1\u03bd\u03ac\u03bb\u03b7\u03c8\u03b7 \u03b5\u03bd\u03cc\u03c2 \u03b9\u03b4\u03b9\u03c9\u03c4\u03b9\u03ba\u03bf\u03cd marker \u03c0\u03bf\u03c5 \u03b5\u03af\u03c7\u03b5 \u03b4\u03bf\u03b8\u03b5\u03af \u03c3\u03b5 \u03c0\u03c1\u03bf\u03b7\u03b3\u03bf\u03cd\u03bc\u03b5\u03bd\u03b7 \u03c3\u03c5\u03bd\u03b5\u03b4\u03c1\u03af\u03b1. \u0391\u03bd \u03b7 \u03bc\u03bd\u03ae\u03bc\u03b7 \u03b4\u03b5\u03bd \u03c4\u03bf \u03b1\u03bd\u03b1\u03ba\u03c4\u03ae\u03c3\u03b5\u03b9 \u03c3\u03c9\u03c3\u03c4\u03ac, \u03bf \u03ba\u03ce\u03b4\u03b9\u03ba\u03b1\u03c2 \u03b1\u03c0\u03bf\u03c4\u03c5\u03b3\u03c7\u03ac\u03bd\u03b5\u03b9, \u03b1\u03ba\u03cc\u03bc\u03b7 \u03ba\u03b9 \u03b1\u03bd \u03b7 \u03b5\u03be\u03ae\u03b3\u03b7\u03c3\u03b7 \u03c4\u03bf\u03c5 \u03bc\u03bf\u03bd\u03c4\u03ad\u03bb\u03bf\u03c5 \u03b5\u03af\u03bd\u03b1\u03b9 \u03c0\u03b5\u03b9\u03c3\u03c4\u03b9\u03ba\u03ae.<\/p>\n<p>\u0397 \u03bb\u03bf\u03b3\u03b9\u03ba\u03ae \u03b1\u03c5\u03c4\u03ae \u03c3\u03c5\u03bd\u03b4\u03ad\u03b5\u03b9 \u03c4\u03b7 \u03bc\u03bd\u03ae\u03bc\u03b7 \u03bc\u03b5 \u03b5\u03c0\u03b9\u03c7\u03b5\u03b9\u03c1\u03b7\u03c3\u03b9\u03b1\u03ba\u03cc \u03b1\u03c0\u03bf\u03c4\u03ad\u03bb\u03b5\u03c3\u03bc\u03b1. \u0393\u03b9\u03b1 \u03bc\u03b9\u03b1 \u03bf\u03bc\u03ac\u03b4\u03b1 \u03c0\u03bf\u03c5 \u03b1\u03bd\u03b1\u03c0\u03c4\u03cd\u03c3\u03c3\u03b5\u03b9 agents, \u03c4\u03bf \u03b5\u03c1\u03ce\u03c4\u03b7\u03bc\u03b1 \u03b4\u03b5\u03bd \u03b5\u03af\u03bd\u03b1\u03b9 \u03bc\u03cc\u03bd\u03bf \u00ab\u03b2\u03c1\u03ae\u03ba\u03b5 \u03c3\u03c7\u03b5\u03c4\u03b9\u03ba\u03cc \u03b1\u03c0\u03cc\u03c3\u03c0\u03b1\u03c3\u03bc\u03b1;\u00bb, \u03b1\u03bb\u03bb\u03ac \u00ab\u03c7\u03c1\u03b7\u03c3\u03b9\u03bc\u03bf\u03c0\u03bf\u03af\u03b7\u03c3\u03b5 \u03c4\u03b7\u03bd \u03ba\u03b1\u03c4\u03ac\u03bb\u03bb\u03b7\u03bb\u03b7 \u03b1\u03c0\u03cc\u03b4\u03b5\u03b9\u03be\u03b7 \u03ba\u03b1\u03b9 \u03c0\u03b1\u03c1\u03ae\u03b3\u03b1\u03b3\u03b5 \u03b1\u03bb\u03bb\u03b1\u03b3\u03ae \u03c0\u03bf\u03c5 \u03c0\u03ad\u03c1\u03b1\u03c3\u03b5 \u03c4\u03bf\u03bd \u03b1\u03bd\u03b5\u03be\u03ac\u03c1\u03c4\u03b7\u03c4\u03bf \u03ad\u03bb\u03b5\u03b3\u03c7\u03bf;\u00bb. \u0395\u03af\u03bd\u03b1\u03b9 \u03b7 \u03af\u03b4\u03b9\u03b1 \u03b1\u03c1\u03c7\u03ae \u03c0\u03bf\u03c5 \u03ba\u03ac\u03bd\u03b5\u03b9 \u03c4\u03b1 <a href=\"https:\/\/twodots.gr\/ai-agents-behavioral-testing\/\">behavioral tests \u03c4\u03c9\u03bd AI agents<\/a> \u03b9\u03c3\u03c7\u03c5\u03c1\u03cc\u03c4\u03b5\u03c1\u03b1 \u03b1\u03c0\u03cc \u03ad\u03bd\u03b1 \u03bc\u03b5\u03bc\u03bf\u03bd\u03c9\u03bc\u03ad\u03bd\u03bf success score.<\/p>\n<p>\u03a4\u03bf benchmark \u03b5\u03af\u03bd\u03b1\u03b9 \u03c3\u03ba\u03cc\u03c0\u03b9\u03bc\u03b1 \u03c3\u03c4\u03b5\u03bd\u03cc. \u0395\u03c3\u03c4\u03b9\u03ac\u03b6\u03b5\u03b9 \u03c3\u03c4\u03b7 \u03c3\u03c5\u03bd\u03ad\u03c7\u03b5\u03b9\u03b1 software \u03b5\u03c1\u03b3\u03b1\u03c3\u03b9\u03ce\u03bd \u03ba\u03b1\u03b9 \u03c3\u03b5 \u03ba\u03c1\u03c5\u03c6\u03ad\u03c2, \u03bc\u03b7 \u03c3\u03c5\u03bd\u03b1\u03b3\u03ce\u03b3\u03b9\u03bc\u03b5\u03c2 \u03c0\u03bb\u03b7\u03c1\u03bf\u03c6\u03bf\u03c1\u03af\u03b5\u03c2. \u0391\u03c5\u03c4\u03ae \u03b7 \u03c3\u03c4\u03b5\u03bd\u03cc\u03c4\u03b7\u03c4\u03b1 \u03b2\u03bf\u03b7\u03b8\u03ac \u03c4\u03b7\u03bd \u03b1\u03b9\u03c4\u03b9\u03ce\u03b4\u03b7 \u03b5\u03c1\u03bc\u03b7\u03bd\u03b5\u03af\u03b1: \u03cc\u03c4\u03b1\u03bd \u03c4\u03bf oracle \u03b1\u03c0\u03bf\u03c4\u03c5\u03b3\u03c7\u03ac\u03bd\u03b5\u03b9, \u03b7 \u03bf\u03bc\u03ac\u03b4\u03b1 \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b5\u03be\u03b5\u03c4\u03ac\u03c3\u03b5\u03b9 \u03b1\u03bd \u03ad\u03c6\u03c4\u03b1\u03b9\u03be\u03b5 \u03b7 \u03b1\u03c0\u03bf\u03c5\u03c3\u03af\u03b1 \u03bc\u03bd\u03ae\u03bc\u03b7\u03c2, \u03c4\u03bf retrieval, \u03c4\u03bf scope, \u03b7 \u03c0\u03b1\u03bb\u03b1\u03af\u03c9\u03c3\u03b7 \u03ae \u03b7 \u03c7\u03c1\u03ae\u03c3\u03b7 \u03c4\u03b7\u03c2 \u03b1\u03bd\u03b1\u03ba\u03c4\u03b7\u03bc\u03ad\u03bd\u03b7\u03c2 \u03b1\u03c0\u03cc\u03b4\u03b5\u03b9\u03be\u03b7\u03c2.<\/p>\n<h2 id=\"pagides-mnimonikis-ygieinis\">\u039f\u03b9 \u03c0\u03b1\u03b3\u03af\u03b4\u03b5\u03c2 \u03c0\u03bf\u03c5 \u03b1\u03c0\u03bf\u03ba\u03b1\u03bb\u03cd\u03c0\u03c4\u03bf\u03c5\u03bd \u03ba\u03b1\u03ba\u03ae \u03bc\u03bd\u03b7\u03bc\u03bf\u03bd\u03b9\u03ba\u03ae \u03c5\u03b3\u03b9\u03b5\u03b9\u03bd\u03ae<\/h2>\n<p>\u03a4\u03bf DreamBench-SWE \u03bf\u03c1\u03b3\u03b1\u03bd\u03ce\u03bd\u03b5\u03b9 \u03c0\u03b1\u03b8\u03bf\u03bb\u03bf\u03b3\u03af\u03b5\u03c2 \u03b5\u03b9\u03b4\u03b9\u03ba\u03ac \u03b3\u03b9\u03b1 software engineering. \u03a0\u03b5\u03c1\u03b9\u03bb\u03b1\u03bc\u03b2\u03ac\u03bd\u03b5\u03b9 \u03c0\u03b1\u03c1\u03c9\u03c7\u03b7\u03bc\u03ad\u03bd\u03b1 \u03b1\u03c1\u03c7\u03b9\u03c4\u03b5\u03ba\u03c4\u03bf\u03bd\u03b9\u03ba\u03ac \u03b4\u03b5\u03b4\u03bf\u03bc\u03ad\u03bd\u03b1, \u03cc\u03c1\u03b9\u03b1 generated files, feedback \u03c0\u03b5\u03c1\u03b9\u03bf\u03c1\u03b9\u03c3\u03bc\u03ad\u03bd\u03bf \u03c3\u03b5 \u03c3\u03c5\u03b3\u03ba\u03b5\u03ba\u03c1\u03b9\u03bc\u03ad\u03bd\u03bf scope, \u03c8\u03b5\u03c5\u03b4\u03ae \u03c3\u03c5\u03bc\u03c0\u03b5\u03c1\u03ac\u03c3\u03bc\u03b1\u03c4\u03b1 \u03b1\u03c0\u03cc failures \u03ba\u03b1\u03b9 \u03c0\u03b5\u03c1\u03b9\u03c0\u03c4\u03ce\u03c3\u03b5\u03b9\u03c2 \u03cc\u03c0\u03bf\u03c5 \u03bf agent \u03c0\u03c1\u03ad\u03c0\u03b5\u03b9 \u03bd\u03b1 \u03b1\u03c0\u03ad\u03c7\u03b5\u03b9 \u03b5\u03c0\u03b5\u03b9\u03b4\u03ae \u03b4\u03b5\u03bd \u03ad\u03c7\u03b5\u03b9 \u03b5\u03c0\u03b1\u03c1\u03ba\u03ae \u03b1\u03c0\u03cc\u03b4\u03b5\u03b9\u03be\u03b7. \u0391\u03c5\u03c4\u03ad\u03c2 \u03bf\u03b9 \u03c0\u03b1\u03b3\u03af\u03b4\u03b5\u03c2 \u03bc\u03bf\u03b9\u03ac\u03b6\u03bf\u03c5\u03bd \u03bc\u03b5 \u03b1\u03c0\u03bf\u03c6\u03ac\u03c3\u03b5\u03b9\u03c2 \u03c0\u03bf\u03c5 \u03b5\u03bc\u03c6\u03b1\u03bd\u03af\u03b6\u03bf\u03bd\u03c4\u03b1\u03b9 \u03c3\u03b5 \u03c0\u03c1\u03b1\u03b3\u03bc\u03b1\u03c4\u03b9\u03ba\u03ac \u03c4\u03b5\u03c7\u03bd\u03b9\u03ba\u03ac workflows.<\/p>\n<ul>\n<li><strong>Stale use:<\/strong> \u03bf agent \u03c7\u03c1\u03b7\u03c3\u03b9\u03bc\u03bf\u03c0\u03bf\u03b9\u03b5\u03af \u03c0\u03bb\u03b7\u03c1\u03bf\u03c6\u03bf\u03c1\u03af\u03b1 \u03c0\u03bf\u03c5 \u03ad\u03c7\u03b5\u03b9 \u03b1\u03bd\u03c4\u03b9\u03ba\u03b1\u03c4\u03b1\u03c3\u03c4\u03b1\u03b8\u03b5\u03af \u03b1\u03c0\u03cc \u03bd\u03b5\u03cc\u03c4\u03b5\u03c1\u03b7 \u03b1\u03c0\u03cc\u03c6\u03b1\u03c3\u03b7.<\/li>\n<li><strong>Overscope:<\/strong> \u03b5\u03c6\u03b1\u03c1\u03bc\u03cc\u03b6\u03b5\u03b9 \u03ad\u03bd\u03b1\u03bd \u03ba\u03b1\u03bd\u03cc\u03bd\u03b1 \u03c0\u03ad\u03c1\u03b1 \u03b1\u03c0\u03cc \u03c4\u03bf module \u03ae \u03c4\u03b7\u03bd \u03c0\u03b5\u03c1\u03af\u03c0\u03c4\u03c9\u03c3\u03b7 \u03cc\u03c0\u03bf\u03c5 \u03b9\u03c3\u03c7\u03cd\u03b5\u03b9.<\/li>\n<li><strong>Repeated error:<\/strong> \u03b5\u03c0\u03b1\u03bd\u03b1\u03bb\u03b1\u03bc\u03b2\u03ac\u03bd\u03b5\u03b9 \u03c0\u03c1\u03bf\u03b7\u03b3\u03bf\u03cd\u03bc\u03b5\u03bd\u03bf \u03bb\u03ac\u03b8\u03bf\u03c2 \u03b5\u03c0\u03b5\u03b9\u03b4\u03ae \u03c4\u03bf \u03b1\u03c0\u03bf\u03b8\u03ae\u03ba\u03b5\u03c5\u03c3\u03b5 \u03c9\u03c2 \u03c7\u03c1\u03ae\u03c3\u03b9\u03bc\u03bf \u03bc\u03ac\u03b8\u03b7\u03bc\u03b1.<\/li>\n<li><strong>Irrelevant import:<\/strong> \u03b5\u03b9\u03c3\u03ac\u03b3\u03b5\u03b9 \u03bc\u03bd\u03ae\u03bc\u03b7 \u03c0\u03bf\u03c5 \u03bc\u03bf\u03b9\u03ac\u03b6\u03b5\u03b9 \u03c3\u03c7\u03b5\u03c4\u03b9\u03ba\u03ae \u03b1\u03bb\u03bb\u03ac \u03b4\u03b5\u03bd \u03c0\u03c1\u03ad\u03c0\u03b5\u03b9 \u03bd\u03b1 \u03b5\u03c0\u03b7\u03c1\u03b5\u03ac\u03c3\u03b5\u03b9 \u03c4\u03b7\u03bd \u03b5\u03c1\u03b3\u03b1\u03c3\u03af\u03b1.<\/li>\n<\/ul>\n<p>\u03a3\u03c4\u03bf v2 \u03ba\u03b1\u03c4\u03b1\u03b3\u03c1\u03ac\u03c6\u03b7\u03ba\u03b1\u03bd IrrelevantImportRate 1434\/3024, OverscopeRate 378\/2268, RepeatedErrorRate 1512\/2646 \u03ba\u03b1\u03b9 StaleUseRate 756\/3780. \u039f\u03b9 \u03b1\u03c1\u03b9\u03b8\u03bc\u03bf\u03af \u03b4\u03b5\u03bd \u03b5\u03af\u03bd\u03b1\u03b9 \u03b3\u03b5\u03bd\u03b9\u03ba\u03ac \u03c0\u03bf\u03c3\u03bf\u03c3\u03c4\u03ac \u03b1\u03c0\u03bf\u03c4\u03c5\u03c7\u03af\u03b1\u03c2 \u03cc\u03bb\u03c9\u03bd \u03c4\u03c9\u03bd agents. \u0391\u03c6\u03bf\u03c1\u03bf\u03cd\u03bd \u03c4\u03bf \u03c3\u03c5\u03b3\u03ba\u03b5\u03ba\u03c1\u03b9\u03bc\u03ad\u03bd\u03bf \u03c0\u03b5\u03b9\u03c1\u03b1\u03bc\u03b1\u03c4\u03b9\u03ba\u03cc \u03c3\u03cd\u03bd\u03bf\u03bb\u03bf \u03ba\u03b1\u03b9 \u03c7\u03c1\u03b7\u03c3\u03b9\u03bc\u03b5\u03cd\u03bf\u03c5\u03bd \u03b3\u03b9\u03b1 \u03bd\u03b1 \u03b5\u03be\u03b7\u03b3\u03ae\u03c3\u03bf\u03c5\u03bd \u03c0\u03bf\u03cd \u03c3\u03c0\u03ac\u03b5\u03b9 \u03ba\u03ac\u03b8\u03b5 \u03c3\u03c5\u03bd\u03b8\u03ae\u03ba\u03b7.<\/p>\n<p>\u0397 \u03b4\u03b9\u03ac\u03ba\u03c1\u03b9\u03c3\u03b7 \u03bc\u03b5\u03c4\u03b1\u03be\u03cd \u03c3\u03c7\u03b5\u03c4\u03b9\u03ba\u03cc\u03c4\u03b7\u03c4\u03b1\u03c2 \u03ba\u03b1\u03b9 \u03b5\u03b3\u03ba\u03c5\u03c1\u03cc\u03c4\u03b7\u03c4\u03b1\u03c2 \u03b5\u03af\u03bd\u03b1\u03b9 \u03b9\u03b4\u03b9\u03b1\u03af\u03c4\u03b5\u03c1\u03b1 \u03c3\u03b7\u03bc\u03b1\u03bd\u03c4\u03b9\u03ba\u03ae. \u0388\u03bd\u03b1 embedding \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b2\u03c1\u03b5\u03b9 \u03c4\u03bf \u03c0\u03b9\u03bf \u03cc\u03bc\u03bf\u03b9\u03bf \u03b1\u03c0\u03cc\u03c3\u03c0\u03b1\u03c3\u03bc\u03b1, \u03b1\u03bb\u03bb\u03ac \u03b7 \u03bf\u03bc\u03bf\u03b9\u03cc\u03c4\u03b7\u03c4\u03b1 \u03b4\u03b5\u03bd \u03b1\u03c0\u03bf\u03b4\u03b5\u03b9\u03ba\u03bd\u03cd\u03b5\u03b9 \u03cc\u03c4\u03b9 \u03c4\u03bf \u03b1\u03c0\u03cc\u03c3\u03c0\u03b1\u03c3\u03bc\u03b1 \u03b5\u03af\u03bd\u03b1\u03b9 \u03c0\u03c1\u03cc\u03c3\u03c6\u03b1\u03c4\u03bf, \u03b5\u03be\u03bf\u03c5\u03c3\u03b9\u03bf\u03b4\u03bf\u03c4\u03b7\u03bc\u03ad\u03bd\u03bf \u03ae \u03b5\u03c6\u03b1\u03c1\u03bc\u03cc\u03c3\u03b9\u03bc\u03bf \u03c3\u03c4\u03bf \u03c4\u03c1\u03ad\u03c7\u03bf\u03bd scope. \u0388\u03bd\u03b1 <a href=\"https:\/\/twodots.gr\/hybrid-ai-search-papers-with-code\/\">\u03c5\u03b2\u03c1\u03b9\u03b4\u03b9\u03ba\u03cc AI search \u03bc\u03b5 \u03bb\u03ad\u03be\u03b5\u03b9\u03c2, embeddings \u03ba\u03b1\u03b9 fallbacks<\/a> \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b2\u03b5\u03bb\u03c4\u03b9\u03ce\u03c3\u03b5\u03b9 \u03c4\u03b7\u03bd \u03b1\u03bd\u03ac\u03ba\u03c4\u03b7\u03c3\u03b7, \u03cc\u03bc\u03c9\u03c2 \u03b7 \u03c4\u03b5\u03bb\u03b9\u03ba\u03ae \u03b1\u03c0\u03cc\u03c6\u03b1\u03c3\u03b7 \u03c7\u03c1\u03b5\u03b9\u03ac\u03b6\u03b5\u03c4\u03b1\u03b9 policy signals \u03c0\u03ad\u03c1\u03b1 \u03b1\u03c0\u03cc similarity.<\/p>\n<h2 id=\"raw-typed-retrieval-gate\">Raw evidence, typed memories \u03ba\u03b1\u03b9 retrieval gate<\/h2>\n<p>\u039f\u03b9 \u03b5\u03c1\u03b5\u03c5\u03bd\u03b7\u03c4\u03ad\u03c2 \u03c7\u03c1\u03b7\u03c3\u03b9\u03bc\u03bf\u03c0\u03bf\u03b9\u03bf\u03cd\u03bd reference probe \u03bc\u03b5 \u03ba\u03cd\u03ba\u03bb\u03bf wake, sleep \u03ba\u03b1\u03b9 read. \u039f \u03cc\u03c1\u03bf\u03c2 \u00absleep\u00bb \u03c0\u03b5\u03c1\u03b9\u03b3\u03c1\u03ac\u03c6\u03b5\u03b9 offline \u03b5\u03c0\u03b5\u03be\u03b5\u03c1\u03b3\u03b1\u03c3\u03af\u03b1 \u03c0\u03ac\u03bd\u03c9 \u03c3\u03c4\u03b1 logs \u03ba\u03b1\u03b9 \u03c3\u03c4\u03b7 \u03bc\u03bd\u03ae\u03bc\u03b7, \u03cc\u03c7\u03b9 \u03b2\u03b9\u03bf\u03bb\u03bf\u03b3\u03b9\u03ba\u03ae \u03b1\u03bd\u03b1\u03bb\u03bf\u03b3\u03af\u03b1. \u0397 \u03b2\u03b1\u03c3\u03b9\u03ba\u03ae \u03b1\u03c1\u03c7\u03ae \u03b5\u03af\u03bd\u03b1\u03b9 \u03cc\u03c4\u03b9 \u03c4\u03b1 raw episodes \u03b4\u03b5\u03bd \u03ba\u03b1\u03c4\u03b1\u03c3\u03c4\u03c1\u03ad\u03c6\u03bf\u03bd\u03c4\u03b1\u03b9. Prompts, \u03ba\u03b1\u03c4\u03ac\u03c3\u03c4\u03b1\u03c3\u03b7 repository, tool calls, outputs, diffs, feedback, memory reads \u03ba\u03b1\u03b9 \u03b1\u03c0\u03bf\u03c4\u03b5\u03bb\u03ad\u03c3\u03bc\u03b1\u03c4\u03b1 \u03c0\u03b1\u03c1\u03b1\u03bc\u03ad\u03bd\u03bf\u03c5\u03bd \u03c9\u03c2 \u03b5\u03bb\u03ad\u03b3\u03be\u03b9\u03bc\u03b7 \u03b1\u03c0\u03cc\u03b4\u03b5\u03b9\u03be\u03b7.<\/p>\n<p>\u03a0\u03ac\u03bd\u03c9 \u03c3\u03b5 \u03b1\u03c5\u03c4\u03ae \u03c4\u03b7 \u03b2\u03ac\u03c3\u03b7 \u03b4\u03b7\u03bc\u03b9\u03bf\u03c5\u03c1\u03b3\u03bf\u03cd\u03bd\u03c4\u03b1\u03b9 typed memories \u03bc\u03b5 provenance. \u039c\u03b9\u03b1 derived \u03bc\u03bd\u03ae\u03bc\u03b7 \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b5\u03c0\u03b9\u03c3\u03b7\u03bc\u03b1\u03bd\u03b8\u03b5\u03af \u03c9\u03c2 stale, \u03bd\u03b1 \u03b1\u03bd\u03c4\u03b9\u03ba\u03b1\u03c4\u03b1\u03c3\u03c4\u03b1\u03b8\u03b5\u03af \u03ae \u03bd\u03b1 \u03b1\u03c0\u03bf\u03ba\u03bb\u03b5\u03b9\u03c3\u03c4\u03b5\u03af \u03b1\u03c0\u03cc \u03c4\u03b7\u03bd \u03b1\u03bd\u03ac\u03ba\u03c4\u03b7\u03c3\u03b7 \u03c7\u03c9\u03c1\u03af\u03c2 \u03bd\u03b1 \u03b4\u03b9\u03b1\u03b3\u03c1\u03ac\u03c6\u03b5\u03c4\u03b1\u03b9 \u03c4\u03bf \u03b1\u03c1\u03c7\u03b9\u03ba\u03cc \u03b5\u03c0\u03b5\u03b9\u03c3\u03cc\u03b4\u03b9\u03bf. \u038c\u03c4\u03b1\u03bd \u03bc\u03b9\u03b1 \u03c3\u03cd\u03bd\u03bf\u03c8\u03b7 \u03b1\u03c0\u03bf\u03b4\u03b5\u03b9\u03c7\u03b8\u03b5\u03af \u03bb\u03b1\u03bd\u03b8\u03b1\u03c3\u03bc\u03ad\u03bd\u03b7, \u03b7 \u03bf\u03bc\u03ac\u03b4\u03b1 \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b5\u03c0\u03b9\u03c3\u03c4\u03c1\u03ad\u03c8\u03b5\u03b9 \u03c3\u03c4\u03b7\u03bd \u03c0\u03c1\u03c9\u03c4\u03bf\u03b3\u03b5\u03bd\u03ae \u03b1\u03c0\u03cc\u03b4\u03b5\u03b9\u03be\u03b7 \u03b1\u03bd\u03c4\u03af \u03bd\u03b1 \u03b5\u03be\u03b1\u03c1\u03c4\u03ac\u03c4\u03b1\u03b9 \u03b1\u03c0\u03cc \u03bc\u03b9\u03b1 \u03b1\u03b4\u03b9\u03b1\u03c6\u03b1\u03bd\u03ae \u03c0\u03b5\u03c1\u03af\u03bb\u03b7\u03c8\u03b7.<\/p>\n<div class=\"td-comparison td-comparison-cards td-comparison-cards--horizontal\">\n<div class=\"td-comparison-grid td-comparison-grid--three\">\n<div class=\"td-platform-card\">\n<p class=\"td-comparison-title\">Raw episode store<\/p>\n<p>\u039a\u03c1\u03b1\u03c4\u03ac \u03c4\u03bf \u03b1\u03c1\u03c7\u03b9\u03ba\u03cc prompt, \u03c4\u03b1 tool calls, \u03c4\u03b1 outputs, \u03c4\u03b1 diffs, \u03c4\u03bf feedback \u03ba\u03b1\u03b9 \u03c4\u03bf outcome. \u0395\u03af\u03bd\u03b1\u03b9 \u03b7 \u03b1\u03bc\u03b5\u03c4\u03ac\u03b2\u03bb\u03b7\u03c4\u03b7 \u03b2\u03ac\u03c3\u03b7 \u03b3\u03b9\u03b1 audit \u03ba\u03b1\u03b9 \u03b5\u03c0\u03b1\u03bd\u03ad\u03bb\u03b5\u03b3\u03c7\u03bf.<\/p>\n<div class=\"td-badge-row\"><span class=\"td-badge\">Evidence<\/span><span class=\"td-badge\">Audit<\/span><\/div>\n<\/div>\n<div class=\"td-platform-card td-platform-card--navy\">\n<p class=\"td-comparison-title\">Typed derived memory<\/p>\n<p>\u039c\u03b5\u03c4\u03b1\u03c4\u03c1\u03ad\u03c0\u03b5\u03b9 \u03b5\u03c0\u03b5\u03b9\u03c3\u03cc\u03b4\u03b9\u03b1 \u03c3\u03b5 facts \u03ae rules \u03bc\u03b5 provenance, scope, \u03c7\u03c1\u03cc\u03bd\u03bf, confidence \u03ba\u03b1\u03b9 status, \u03c7\u03c9\u03c1\u03af\u03c2 \u03bd\u03b1 \u03b1\u03bd\u03c4\u03b9\u03ba\u03b1\u03b8\u03b9\u03c3\u03c4\u03ac \u03c4\u03b7\u03bd \u03c0\u03c1\u03c9\u03c4\u03bf\u03b3\u03b5\u03bd\u03ae \u03c0\u03b7\u03b3\u03ae.<\/p>\n<div class=\"td-badge-row\"><span class=\"td-badge\">Scope<\/span><span class=\"td-badge\">Supersession<\/span><\/div>\n<\/div>\n<div class=\"td-platform-card\">\n<p class=\"td-comparison-title\">Retrieval gate<\/p>\n<p>\u0391\u03c0\u03bf\u03c6\u03b1\u03c3\u03af\u03b6\u03b5\u03b9 \u03c4\u03b9 \u03b5\u03c0\u03b9\u03c4\u03c1\u03ad\u03c0\u03b5\u03c4\u03b1\u03b9 \u03bd\u03b1 \u03c6\u03c4\u03ac\u03c3\u03b5\u03b9 \u03c3\u03c4\u03bf context \u03c4\u03b7\u03c2 \u03c3\u03c5\u03b3\u03ba\u03b5\u03ba\u03c1\u03b9\u03bc\u03ad\u03bd\u03b7\u03c2 \u03b5\u03c1\u03b3\u03b1\u03c3\u03af\u03b1\u03c2 \u03ba\u03b1\u03b9 \u03bc\u03c0\u03bb\u03bf\u03ba\u03ac\u03c1\u03b5\u03b9 stale, \u03ac\u03c3\u03c7\u03b5\u03c4\u03b7 \u03ae \u03c5\u03c0\u03b5\u03c1\u03b2\u03bf\u03bb\u03b9\u03ba\u03ac \u03b5\u03c5\u03c1\u03b5\u03af\u03b1 \u03bc\u03bd\u03ae\u03bc\u03b7.<\/p>\n<div class=\"td-badge-row\"><span class=\"td-badge\">Policy<\/span><span class=\"td-badge\">Context<\/span><\/div>\n<\/div>\n<\/div>\n<\/div>\n<p>\u03a4\u03bf retrieval gate \u03b4\u03b5\u03bd \u03b2\u03b5\u03bb\u03c4\u03b9\u03c3\u03c4\u03bf\u03c0\u03bf\u03b9\u03b5\u03af \u03b1\u03c0\u03bb\u03ce\u03c2 \u03c4\u03b7 \u03bc\u03ad\u03b3\u03b9\u03c3\u03c4\u03b7 \u03b1\u03bd\u03ac\u03ba\u03bb\u03b7\u03c3\u03b7. \u0395\u03be\u03b9\u03c3\u03bf\u03c1\u03c1\u03bf\u03c0\u03b5\u03af \u03c7\u03c1\u03ae\u03c3\u03b9\u03bc\u03b7 \u03bc\u03bd\u03ae\u03bc\u03b7 \u03ba\u03b1\u03b9 \u03b5\u03c0\u03b9\u03b2\u03bb\u03b1\u03b2\u03ae \u03b5\u03b9\u03c3\u03b1\u03b3\u03c9\u03b3\u03ae. \u03a3\u03c4\u03b7\u03bd \u03c0\u03c1\u03ac\u03be\u03b7, \u03bc\u03b9\u03b1 \u03ce\u03c1\u03b9\u03bc\u03b7 \u03bc\u03bd\u03ae\u03bc\u03b7 \u03c0\u03c1\u03ad\u03c0\u03b5\u03b9 \u03bd\u03b1 \u03b3\u03bd\u03c9\u03c1\u03af\u03b6\u03b5\u03b9 \u03cc\u03c7\u03b9 \u03bc\u03cc\u03bd\u03bf \u03c4\u03b9 \u03bd\u03b1 \u03b5\u03bc\u03c6\u03b1\u03bd\u03af\u03c3\u03b5\u03b9, \u03b1\u03bb\u03bb\u03ac \u03ba\u03b1\u03b9 \u03c4\u03b9 \u03bd\u03b1 \u03ba\u03c1\u03b1\u03c4\u03ae\u03c3\u03b5\u03b9 \u03ad\u03be\u03c9 \u03b1\u03c0\u03cc \u03c4\u03bf context. \u0397 \u03b1\u03c0\u03cc\u03c6\u03b1\u03c3\u03b7 \u03b1\u03c5\u03c4\u03ae \u03b1\u03bd\u03ae\u03ba\u03b5\u03b9 \u03c3\u03c4\u03bf harness \u03ba\u03b1\u03b9 \u03c0\u03c1\u03ad\u03c0\u03b5\u03b9 \u03bd\u03b1 \u03b5\u03af\u03bd\u03b1\u03b9 \u03c0\u03b1\u03c1\u03b1\u03c4\u03b7\u03c1\u03ae\u03c3\u03b9\u03bc\u03b7, \u03cc\u03c0\u03c9\u03c2 \u03ba\u03b1\u03b9 \u03c4\u03b1 \u03c5\u03c0\u03cc\u03bb\u03bf\u03b9\u03c0\u03b1 \u03b5\u03c0\u03af\u03c0\u03b5\u03b4\u03b1 \u03b5\u03bd\u03cc\u03c2 <a href=\"https:\/\/twodots.gr\/ai-agent-harnesses-elegchos-kostos-self-hosting\/\">AI agent harness \u03bc\u03b5 \u03b5\u03bb\u03b5\u03b3\u03c7\u03cc\u03bc\u03b5\u03bd\u03bf \u03ba\u03cc\u03c3\u03c4\u03bf\u03c2 \u03ba\u03b1\u03b9 \u03c3\u03b1\u03c6\u03ae \u03cc\u03c1\u03b9\u03b1<\/a>.<\/p>\n<h2 id=\"apotelesmata-v2-v21\">\u03a4\u03b9 \u03ad\u03b4\u03b5\u03b9\u03be\u03b1\u03bd \u03c4\u03b1 v2 \u03ba\u03b1\u03b9 v2.1<\/h2>\n<p>\u03a3\u03c4\u03bf scaled v2 fold, \u03c4\u03bf typed-plus-raw reference probe \u03c0\u03ad\u03c4\u03c5\u03c7\u03b5 95 \u03b1\u03c0\u03cc 180 \u03b4\u03bf\u03ba\u03b9\u03bc\u03ad\u03c2, \u03b5\u03bd\u03ce \u03c4\u03bf \u03b9\u03c3\u03c7\u03c5\u03c1\u03cc verbatim event memory \u03c0\u03ad\u03c4\u03c5\u03c7\u03b5 89 \u03b1\u03c0\u03cc 180. \u0397 clustered \u03b1\u03bd\u03ac\u03bb\u03c5\u03c3\u03b7 \u03ad\u03b4\u03c9\u03c3\u03b5 p=0,518 \u03ba\u03b1\u03b9 Holm-adjusted p=1. \u0397 \u03b1\u03c1\u03b9\u03b8\u03bc\u03b7\u03c4\u03b9\u03ba\u03ae \u03b4\u03b9\u03b1\u03c6\u03bf\u03c1\u03ac \u03ad\u03be\u03b9 \u03b5\u03c0\u03b9\u03c4\u03c5\u03c7\u03b9\u03ce\u03bd \u03b4\u03b5\u03bd \u03c4\u03b5\u03ba\u03bc\u03b7\u03c1\u03b9\u03ce\u03bd\u03b5\u03b9 \u03c5\u03c0\u03b5\u03c1\u03bf\u03c7\u03ae, \u03b5\u03bd\u03ce \u03b7 \u03bc\u03b7 \u03b1\u03c0\u03cc\u03c1\u03c1\u03b9\u03c8\u03b7 \u03c4\u03b7\u03c2 \u03bc\u03b7\u03b4\u03b5\u03bd\u03b9\u03ba\u03ae\u03c2 \u03c5\u03c0\u03cc\u03b8\u03b5\u03c3\u03b7\u03c2 \u03b4\u03b5\u03bd \u03b1\u03c0\u03bf\u03b4\u03b5\u03b9\u03ba\u03bd\u03cd\u03b5\u03b9 \u03b9\u03c3\u03bf\u03b4\u03c5\u03bd\u03b1\u03bc\u03af\u03b1.<\/p>\n<p>\u03a4\u03bf v2 \u03b1\u03c0\u03bf\u03ba\u03ac\u03bb\u03c5\u03c8\u03b5 \u03b5\u03c0\u03af\u03c3\u03b7\u03c2 \u03c0\u03b5\u03c1\u03b9\u03bf\u03c1\u03b9\u03c3\u03bc\u03bf\u03cd\u03c2 \u03c3\u03c4\u03b1 constructs C9 \u03ba\u03b1\u03b9 C10, \u03cc\u03c0\u03bf\u03c5 \u03b7 baseline \u03c7\u03c9\u03c1\u03af\u03c2 \u03bc\u03bd\u03ae\u03bc\u03b7 \u03b4\u03b5\u03bd \u03ac\u03c6\u03b7\u03bd\u03b5 \u03b1\u03c1\u03ba\u03b5\u03c4\u03cc headroom. \u0386\u03c1\u03b1 \u03c4\u03bf benchmark \u03b4\u03b5\u03bd \u03b4\u03b9\u03ba\u03b1\u03b9\u03bf\u03bb\u03bf\u03b3\u03b5\u03af \u03b5\u03c5\u03c1\u03cd \u03b9\u03c3\u03c7\u03c5\u03c1\u03b9\u03c3\u03bc\u03cc \u03cc\u03c4\u03b9 \u03bc\u03b5\u03c4\u03c1\u03ac \u03ba\u03ac\u03b8\u03b5 \u03bc\u03bf\u03c1\u03c6\u03ae \u00abanti-hoarding\u00bb \u03c3\u03c5\u03bc\u03c0\u03b5\u03c1\u03b9\u03c6\u03bf\u03c1\u03ac\u03c2 \u03bc\u03b5 \u03c4\u03bf\u03bd \u03af\u03b4\u03b9\u03bf \u03b2\u03b1\u03b8\u03bc\u03cc \u03b5\u03b3\u03ba\u03c5\u03c1\u03cc\u03c4\u03b7\u03c4\u03b1\u03c2.<\/p>\n<p>\u03a4\u03bf v2.1 \u03c3\u03c7\u03b5\u03b4\u03b9\u03ac\u03c3\u03c4\u03b7\u03ba\u03b5 \u03c9\u03c2 \u03c7\u03c9\u03c1\u03b9\u03c3\u03c4\u03cc preregistered successor audit \u03ba\u03b1\u03b9 \u03c0\u03ac\u03b3\u03c9\u03c3\u03b5 \u03c0\u03c1\u03b9\u03bd \u03b1\u03c0\u03cc \u03c4\u03b7\u03bd \u03b5\u03c0\u03b9\u03b8\u03b5\u03ce\u03c1\u03b7\u03c3\u03b7 \u03c4\u03c9\u03bd \u03b1\u03c0\u03bf\u03c4\u03b5\u03bb\u03b5\u03c3\u03bc\u03ac\u03c4\u03c9\u03bd. \u039f\u03bb\u03bf\u03ba\u03bb\u03ae\u03c1\u03c9\u03c3\u03b5 360\/360 work units \u03ba\u03b1\u03b9 720\/720 S3 cells \u03c3\u03b5 \u03c4\u03ad\u03c3\u03c3\u03b5\u03c1\u03b9\u03c2 \u03c3\u03c5\u03bd\u03b8\u03ae\u03ba\u03b5\u03c2.<\/p>\n<div class=\"td-chart td-chart--metrics\">\n<div class=\"td-chart-head\">\n<p class=\"td-chart-title\">S3 Pass@1 \u03c3\u03c4\u03bf \u03c7\u03c9\u03c1\u03b9\u03c3\u03c4\u03cc v2.1 successor audit<\/p>\n<p class=\"td-chart-subtitle\">\u0391\u03ba\u03c1\u03b9\u03b2\u03ae \u03b1\u03c0\u03bf\u03c4\u03b5\u03bb\u03ad\u03c3\u03bc\u03b1\u03c4\u03b1 \u03c4\u03c9\u03bd \u03c4\u03b5\u03c3\u03c3\u03ac\u03c1\u03c9\u03bd admissible conditions\u00b7 \u03cc\u03c7\u03b9 \u03b3\u03b5\u03bd\u03b9\u03ba\u03ae \u03ba\u03b1\u03c4\u03ac\u03c4\u03b1\u03be\u03b7 \u03c0\u03c1\u03bf\u03ca\u03cc\u03bd\u03c4\u03c9\u03bd \u03ae \u03b1\u03c1\u03c7\u03b9\u03c4\u03b5\u03ba\u03c4\u03bf\u03bd\u03b9\u03ba\u03ce\u03bd \u03bc\u03bd\u03ae\u03bc\u03b7\u03c2.<\/p>\n<\/div>\n<div class=\"td-chart-body\">\n<div class=\"td-metric-grid\">\n<div class=\"td-metric-card\"><span class=\"td-metric-value\">21\/180<\/span><span class=\"td-metric-label\">\u03c7\u03c9\u03c1\u03af\u03c2 \u03b5\u03be\u03c9\u03c4\u03b5\u03c1\u03b9\u03ba\u03ae \u03bc\u03bd\u03ae\u03bc\u03b7<\/span><span class=\"td-metric-note\">B0 baseline<\/span><\/div>\n<div class=\"td-metric-card\"><span class=\"td-metric-value\">82\/180<\/span><span class=\"td-metric-label\">verbatim event memory<\/span><span class=\"td-metric-note\">Deterministic control<\/span><\/div>\n<div class=\"td-metric-card\"><span class=\"td-metric-value\">83\/180<\/span><span class=\"td-metric-label\">typed plus raw probe<\/span><span class=\"td-metric-note\">Reference architecture<\/span><\/div>\n<div class=\"td-metric-card\"><span class=\"td-metric-value\">97\/180<\/span><span class=\"td-metric-label\">pinned Mem0 literal storage<\/span><span class=\"td-metric-note\">\u039c\u03af\u03b1 \u03b1\u03ba\u03c1\u03b9\u03b2\u03ae\u03c2 hosted \u03c1\u03cd\u03b8\u03bc\u03b9\u03c3\u03b7<\/span><\/div>\n<\/div>\n<\/div>\n<\/div>\n<p>\u039f\u03b9 \u03c4\u03c1\u03b5\u03b9\u03c2 \u03b4\u03b9\u03b1\u03b8\u03ad\u03c3\u03b9\u03bc\u03b5\u03c2 \u03c3\u03c5\u03b3\u03ba\u03c1\u03af\u03c3\u03b5\u03b9\u03c2 \u03b1\u03c0\u03ad\u03bd\u03b1\u03bd\u03c4\u03b9 \u03c3\u03c4\u03b7 \u03c3\u03c5\u03bd\u03b8\u03ae\u03ba\u03b7 \u03c7\u03c9\u03c1\u03af\u03c2 \u03bc\u03bd\u03ae\u03bc\u03b7 \u03b1\u03c0\u03ad\u03c1\u03c1\u03b9\u03c8\u03b1\u03bd \u03c4\u03b7 \u03bc\u03b7\u03b4\u03b5\u03bd\u03b9\u03ba\u03ae \u03c5\u03c0\u03cc\u03b8\u03b5\u03c3\u03b7 \u03bc\u03b5\u03c4\u03ac \u03c4\u03b7 \u03b4\u03b9\u03cc\u03c1\u03b8\u03c9\u03c3\u03b7 Holm. \u0391\u03c5\u03c4\u03cc \u03c5\u03c0\u03bf\u03c3\u03c4\u03b7\u03c1\u03af\u03b6\u03b5\u03b9 \u03cc\u03c4\u03b9 \u03c4\u03bf benchmark \u03b4\u03b9\u03b1\u03ba\u03c1\u03af\u03bd\u03b5\u03b9 memory-bearing \u03c3\u03c5\u03bd\u03b8\u03ae\u03ba\u03b5\u03c2 \u03b1\u03c0\u03cc \u03c4\u03b7 \u03c3\u03c5\u03b3\u03ba\u03b5\u03ba\u03c1\u03b9\u03bc\u03ad\u03bd\u03b7 baseline \u03ba\u03b1\u03b9 \u03c0\u03b5\u03c1\u03b9\u03b3\u03c1\u03ac\u03c6\u03b5\u03b9 \u03c4\u03b7\u03bd \u03b1\u03c0\u03cc\u03b4\u03bf\u03c3\u03b7 \u03c4\u03b7\u03c2 \u03b1\u03ba\u03c1\u03b9\u03b2\u03bf\u03cd\u03c2 hosted \u03c1\u03cd\u03b8\u03bc\u03b9\u03c3\u03b7\u03c2 \u03c0\u03bf\u03c5 \u03b4\u03bf\u03ba\u03b9\u03bc\u03ac\u03c3\u03c4\u03b7\u03ba\u03b5. \u0394\u03b5\u03bd \u03b1\u03c0\u03bf\u03b4\u03b5\u03b9\u03ba\u03bd\u03cd\u03b5\u03b9 \u03cc\u03c4\u03b9 \u03b7 Mem0 \u03b5\u03af\u03bd\u03b1\u03b9 \u03b3\u03b5\u03bd\u03b9\u03ba\u03ac \u03ba\u03b1\u03bb\u03cd\u03c4\u03b5\u03c1\u03b7 \u03bf\u03cd\u03c4\u03b5 \u03cc\u03c4\u03b9 \u03c3\u03c5\u03b3\u03ba\u03b5\u03ba\u03c1\u03b9\u03bc\u03ad\u03bd\u03bf\u03c2 \u03bc\u03b7\u03c7\u03b1\u03bd\u03b9\u03c3\u03bc\u03cc\u03c2 \u03c0\u03c1\u03bf\u03ba\u03ac\u03bb\u03b5\u03c3\u03b5 \u03c4\u03b7 \u03b4\u03b9\u03b1\u03c6\u03bf\u03c1\u03ac.<\/p>\n<p>\u039f\u03b9 preregistered mechanism comparisons \u03b4\u03b5\u03bd \u03ae\u03c4\u03b1\u03bd \u03b4\u03b9\u03b1\u03b8\u03ad\u03c3\u03b9\u03bc\u03b5\u03c2 \u03b5\u03c0\u03b5\u03b9\u03b4\u03ae \u03bf\u03b9 \u03b1\u03bd\u03c4\u03af\u03c3\u03c4\u03bf\u03b9\u03c7\u03b5\u03c2 conditions \u03b1\u03c0\u03bf\u03c1\u03c1\u03af\u03c6\u03b8\u03b7\u03ba\u03b1\u03bd \u03c0\u03c1\u03b9\u03bd \u03b1\u03c0\u03cc \u03c4\u03b7\u03bd \u03b1\u03be\u03b9\u03bf\u03bb\u03cc\u03b3\u03b7\u03c3\u03b7 \u03bb\u03cc\u03b3\u03c9 \u03bc\u03b7 \u03c3\u03c5\u03bc\u03bc\u03cc\u03c1\u03c6\u03c9\u03c3\u03b7\u03c2. \u0397 \u03c3\u03cd\u03b3\u03ba\u03c1\u03b9\u03c3\u03b7 literal storage \u03bc\u03b5 verbatim memory \u03ae\u03c4\u03b1\u03bd \u03bc\u03b7 \u03b5\u03c0\u03b9\u03b2\u03b5\u03b2\u03b1\u03b9\u03c9\u03c4\u03b9\u03ba\u03ae \u03ba\u03b1\u03b9 \u03b5\u03c5\u03b1\u03af\u03c3\u03b8\u03b7\u03c4\u03b7 \u03c3\u03c4\u03b7\u03bd \u03b1\u03bd\u03ac\u03bb\u03c5\u03c3\u03b7, \u03b5\u03bd\u03ce \u03b7 \u03c3\u03cd\u03b3\u03ba\u03c1\u03b9\u03c3\u03b7 \u03bc\u03b5 \u03c4\u03bf reference probe \u03b4\u03b5\u03bd \u03b1\u03c0\u03ad\u03c1\u03c1\u03b9\u03c8\u03b5. \u03a4\u03bf \u03c4\u03af\u03bc\u03b9\u03bf \u03c3\u03c5\u03bc\u03c0\u03ad\u03c1\u03b1\u03c3\u03bc\u03b1 \u03b5\u03af\u03bd\u03b1\u03b9 benchmark discrimination \u03ba\u03b1\u03b9 \u03ad\u03bd\u03b1 \u03c0\u03b5\u03c1\u03b9\u03bf\u03c1\u03b9\u03c3\u03bc\u03ad\u03bd\u03bf external-system profile.<\/p>\n<h2 id=\"peiramatikoi-elegchoi\">\u03a0\u03ce\u03c2 \u03bf\u03b9 \u03c0\u03b5\u03b9\u03c1\u03b1\u03bc\u03b1\u03c4\u03b9\u03ba\u03bf\u03af \u03ad\u03bb\u03b5\u03b3\u03c7\u03bf\u03b9 \u03ba\u03c1\u03b1\u03c4\u03bf\u03cd\u03bd \u03c4\u03bf benchmark \u03c4\u03af\u03bc\u03b9\u03bf<\/h2>\n<p>\u0388\u03bd\u03b1 benchmark \u03bc\u03bd\u03ae\u03bc\u03b7\u03c2 \u03ba\u03b9\u03bd\u03b4\u03c5\u03bd\u03b5\u03cd\u03b5\u03b9 \u03bd\u03b1 \u03bc\u03b5\u03c4\u03c1\u03ae\u03c3\u03b5\u03b9 \u03ba\u03ac\u03c4\u03b9 \u03b4\u03b9\u03b1\u03c6\u03bf\u03c1\u03b5\u03c4\u03b9\u03ba\u03cc \u03b1\u03c0\u03cc \u03b1\u03c5\u03c4\u03cc \u03c0\u03bf\u03c5 \u03b4\u03b7\u03bb\u03ce\u03bd\u03b5\u03b9. \u039f agent \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b2\u03c1\u03b5\u03b9 \u03c4\u03bf \u03bc\u03c5\u03c3\u03c4\u03b9\u03ba\u03cc \u03c3\u03c4\u03bf\u03b9\u03c7\u03b5\u03af\u03bf \u03bc\u03ad\u03c3\u03b1 \u03c3\u03c4\u03bf repository, \u03bd\u03b1 \u03c4\u03bf \u03ad\u03c7\u03b5\u03b9 \u03c3\u03c5\u03bd\u03b1\u03bd\u03c4\u03ae\u03c3\u03b5\u03b9 \u03ba\u03b1\u03c4\u03ac \u03c4\u03b7\u03bd \u03c0\u03c1\u03bf\u03b5\u03ba\u03c0\u03b1\u03af\u03b4\u03b5\u03c5\u03c3\u03b7 \u03ae \u03bd\u03b1 \u03b5\u03c0\u03c9\u03c6\u03b5\u03bb\u03b7\u03b8\u03b5\u03af \u03b1\u03c0\u03cc \u03b4\u03b9\u03b1\u03c1\u03c1\u03bf\u03ae \u03b1\u03bd\u03ac\u03bc\u03b5\u03c3\u03b1 \u03c3\u03c4\u03b9\u03c2 \u03c3\u03c5\u03bd\u03b8\u03ae\u03ba\u03b5\u03c2. \u03a4\u03bf DreamBench-SWE \u03c7\u03c1\u03b7\u03c3\u03b9\u03bc\u03bf\u03c0\u03bf\u03b9\u03b5\u03af \u03ba\u03c1\u03c5\u03c0\u03c4\u03bf\u03b3\u03c1\u03b1\u03c6\u03b9\u03ba\u03ac \u03c4\u03c5\u03c7\u03b1\u03af\u03b1 injected tokens, \u03b1\u03c0\u03bf\u03bc\u03cc\u03bd\u03c9\u03c3\u03b7 \u03c4\u03c9\u03bd scored artifacts \u03b1\u03c0\u03cc \u03c4\u03bf\u03bd wake agent \u03ba\u03b1\u03b9 contamination checks. \u0397 \u03bc\u03b5\u03bb\u03ad\u03c4\u03b7 \u03b4\u03b7\u03bb\u03ce\u03bd\u03b5\u03b9 filesystem isolation, \u03cc\u03c7\u03b9 \u03b1\u03c0\u03cc\u03bb\u03c5\u03c4\u03b7 network isolation.<\/p>\n<p>\u0397 \u03b1\u03bd\u03ac\u03bb\u03c5\u03c3\u03b7 \u03b3\u03af\u03bd\u03b5\u03c4\u03b1\u03b9 \u03c3\u03b5 \u03b5\u03c0\u03af\u03c0\u03b5\u03b4\u03bf trap-cluster \u03ba\u03b1\u03b9 \u03cc\u03c7\u03b9 \u03bc\u03b5 \u03b1\u03c6\u03b5\u03bb\u03ae \u03c3\u03c5\u03b3\u03ba\u03ad\u03bd\u03c4\u03c1\u03c9\u03c3\u03b7 \u03cc\u03bb\u03c9\u03bd \u03c4\u03c9\u03bd cells \u03c3\u03b1\u03bd \u03bd\u03b1 \u03ae\u03c4\u03b1\u03bd \u03b1\u03bd\u03b5\u03be\u03ac\u03c1\u03c4\u03b7\u03c4\u03b1. \u03a3\u03c4\u03bf v2 \u03c7\u03c1\u03b7\u03c3\u03b9\u03bc\u03bf\u03c0\u03bf\u03b9\u03ae\u03b8\u03b7\u03ba\u03b1\u03bd 60 traps, \u03c4\u03c1\u03af\u03b1 seeds \u03ba\u03b1\u03b9 1.890 condition-level result files. \u03a0\u03bf\u03bb\u03bb\u03b1\u03c0\u03bb\u03ac outcomes \u03b1\u03c0\u03cc \u03c4\u03bf \u03af\u03b4\u03b9\u03bf trap \u03bc\u03bf\u03b9\u03c1\u03ac\u03b6\u03bf\u03bd\u03c4\u03b1\u03b9 \u03ba\u03bf\u03b9\u03bd\u03ae \u03b4\u03bf\u03bc\u03ae\u00b7 \u03b1\u03bd \u03bc\u03b5\u03c4\u03c1\u03b7\u03b8\u03bf\u03cd\u03bd \u03c9\u03c2 \u03c0\u03bb\u03ae\u03c1\u03c9\u03c2 \u03b1\u03bd\u03b5\u03be\u03ac\u03c1\u03c4\u03b7\u03c4\u03b1, \u03b7 \u03c3\u03c4\u03b1\u03c4\u03b9\u03c3\u03c4\u03b9\u03ba\u03ae \u03b2\u03b5\u03b2\u03b1\u03b9\u03cc\u03c4\u03b7\u03c4\u03b1 \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03c6\u03b1\u03bd\u03b5\u03af \u03bc\u03b5\u03b3\u03b1\u03bb\u03cd\u03c4\u03b5\u03c1\u03b7 \u03b1\u03c0\u03cc \u03cc\u03c3\u03bf \u03b5\u03af\u03bd\u03b1\u03b9.<\/p>\n<p>\u03a4\u03bf admission funnel \u03b5\u03bb\u03ad\u03b3\u03c7\u03b5\u03b9 \u03b1\u03bd \u03ad\u03bd\u03b1 trap \u03b5\u03af\u03bd\u03b1\u03b9 \u03ad\u03b3\u03ba\u03c5\u03c1\u03bf \u03c0\u03c1\u03b9\u03bd \u03c3\u03c5\u03bc\u03b2\u03ac\u03bb\u03b5\u03b9 \u03c3\u03c4\u03b1 confirmatory claims. \u039f\u03b9 \u03c3\u03c5\u03b3\u03b3\u03c1\u03b1\u03c6\u03b5\u03af\u03c2 \u03b1\u03bd\u03b1\u03c6\u03ad\u03c1\u03bf\u03c5\u03bd 30 authored, 30 dry-valid, 30 live-valid \u03ba\u03b1\u03b9 30 admitted traps \u03c3\u03c4\u03bf \u03c3\u03c7\u03b5\u03c4\u03b9\u03ba\u03cc fold. \u03a3\u03c4\u03bf successor audit, conditions \u03c0\u03bf\u03c5 \u03b4\u03b5\u03bd \u03c0\u03ad\u03c1\u03b1\u03c3\u03b1\u03bd \u03c4\u03b9\u03c2 \u03c0\u03b1\u03b3\u03c9\u03bc\u03ad\u03bd\u03b5\u03c2 \u03c0\u03c1\u03bf\u03cb\u03c0\u03bf\u03b8\u03ad\u03c3\u03b5\u03b9\u03c2 \u03c3\u03c5\u03bc\u03bc\u03cc\u03c1\u03c6\u03c9\u03c3\u03b7\u03c2 \u03b1\u03c0\u03bf\u03ba\u03bb\u03b5\u03af\u03c3\u03c4\u03b7\u03ba\u03b1\u03bd \u03c0\u03c1\u03b9\u03bd \u03b5\u03be\u03b5\u03c4\u03b1\u03c3\u03c4\u03b5\u03af \u03b7 \u03b1\u03c0\u03cc\u03b4\u03bf\u03c3\u03ae \u03c4\u03bf\u03c5\u03c2. \u0388\u03c4\u03c3\u03b9 \u03c0\u03c1\u03bf\u03ad\u03ba\u03c5\u03c8\u03b1\u03bd \u00ab\u03bc\u03b7 \u03b4\u03b9\u03b1\u03b8\u03ad\u03c3\u03b9\u03bc\u03b5\u03c2\u00bb \u03c3\u03c5\u03b3\u03ba\u03c1\u03af\u03c3\u03b5\u03b9\u03c2, \u03b1\u03bd\u03c4\u03af \u03bd\u03b1 \u03b1\u03bb\u03bb\u03ac\u03be\u03bf\u03c5\u03bd \u03bf\u03b9 \u03ba\u03b1\u03bd\u03cc\u03bd\u03b5\u03c2 \u03b1\u03c6\u03bf\u03cd \u03ad\u03b3\u03b9\u03bd\u03b1\u03bd \u03b3\u03bd\u03c9\u03c3\u03c4\u03ac \u03c4\u03b1 \u03b1\u03c0\u03bf\u03c4\u03b5\u03bb\u03ad\u03c3\u03bc\u03b1\u03c4\u03b1.<\/p>\n<aside class=\"td-article-note\">\n<p><strong>\u03a4\u03bf acceptance criterion \u03c0\u03c1\u03ad\u03c0\u03b5\u03b9 \u03bd\u03b1 \u03c0\u03b1\u03b3\u03ce\u03bd\u03b5\u03b9 \u03c0\u03c1\u03b9\u03bd \u03b1\u03c0\u03cc \u03c4\u03bf \u03b1\u03c0\u03bf\u03c4\u03ad\u03bb\u03b5\u03c3\u03bc\u03b1.<\/strong> \u0391\u03bd \u03bc\u03b9\u03b1 \u03b5\u03c4\u03b1\u03b9\u03c1\u03b9\u03ba\u03ae \u03bf\u03bc\u03ac\u03b4\u03b1 \u03b1\u03bb\u03bb\u03ac\u03be\u03b5\u03b9 \u03c4\u03bf oracle \u03ae \u03b1\u03c6\u03b1\u03b9\u03c1\u03ad\u03c3\u03b5\u03b9 \u03b4\u03cd\u03c3\u03ba\u03bf\u03bb\u03b5\u03c2 \u03c0\u03b5\u03c1\u03b9\u03c0\u03c4\u03ce\u03c3\u03b5\u03b9\u03c2 \u03b1\u03c6\u03bf\u03cd \u03b4\u03b5\u03b9 \u03c0\u03bf\u03b9\u03b1 \u03bb\u03cd\u03c3\u03b7 \u03ba\u03b5\u03c1\u03b4\u03af\u03b6\u03b5\u03b9, \u03c4\u03bf pilot \u03c0\u03b1\u03cd\u03b5\u03b9 \u03bd\u03b1 \u03b5\u03af\u03bd\u03b1\u03b9 \u03b1\u03be\u03b9\u03bf\u03bb\u03cc\u03b3\u03b7\u03c3\u03b7 \u03ba\u03b1\u03b9 \u03b3\u03af\u03bd\u03b5\u03c4\u03b1\u03b9 \u03b5\u03ba \u03c4\u03c9\u03bd \u03c5\u03c3\u03c4\u03ad\u03c1\u03c9\u03bd \u03b4\u03b9\u03ba\u03b1\u03b9\u03bf\u03bb\u03cc\u03b3\u03b7\u03c3\u03b7 \u03b1\u03b3\u03bf\u03c1\u03ac\u03c2.<\/p>\n<\/aside>\n<p>\u0393\u03b9\u03b1 \u03bc\u03b9\u03b1 \u03b5\u03c4\u03b1\u03b9\u03c1\u03b9\u03ba\u03ae \u03bf\u03bc\u03ac\u03b4\u03b1, \u03b1\u03c5\u03c4\u03cc \u03bc\u03b5\u03c4\u03b1\u03c6\u03c1\u03ac\u03b6\u03b5\u03c4\u03b1\u03b9 \u03c3\u03b5 \u03c4\u03c1\u03b5\u03b9\u03c2 \u03b1\u03c1\u03c7\u03ad\u03c2: \u03ad\u03b3\u03ba\u03c1\u03b9\u03c3\u03b7 test cases \u03c0\u03c1\u03b9\u03bd \u03b1\u03c0\u03cc \u03c4\u03bf run, \u03b1\u03bc\u03b5\u03c4\u03ac\u03b2\u03bb\u03b7\u03c4\u03b1 acceptance criteria \u03ba\u03b1\u03b9 \u03ba\u03b1\u03c4\u03b1\u03b3\u03c1\u03b1\u03c6\u03ae \u03c4\u03c9\u03bd \u03ac\u03ba\u03c5\u03c1\u03c9\u03bd \u03ae \u03bc\u03b7 \u03c3\u03c5\u03bc\u03bc\u03bf\u03c1\u03c6\u03bf\u03cd\u03bc\u03b5\u03bd\u03c9\u03bd conditions. \u0397 \u03c0\u03c1\u03b1\u03ba\u03c4\u03b9\u03ba\u03ae \u03c3\u03c5\u03bc\u03c0\u03bb\u03b7\u03c1\u03ce\u03bd\u03b5\u03b9 \u03c4\u03bf\u03bd \u03ad\u03bb\u03b5\u03b3\u03c7\u03bf attribution: \u03cc\u03c4\u03b1\u03bd \u03ad\u03bd\u03b1\u03c2 agent \u03b1\u03c0\u03bf\u03c4\u03c5\u03b3\u03c7\u03ac\u03bd\u03b5\u03b9, \u03c7\u03c1\u03b5\u03b9\u03ac\u03b6\u03b5\u03c4\u03b1\u03b9 \u03bd\u03b1 \u03be\u03b5\u03c7\u03c9\u03c1\u03af\u03c3\u03bf\u03c5\u03bc\u03b5 <a href=\"https:\/\/twodots.gr\/ai-agents-poios-ftaiei-otan-aftomatopoiisi-apotygchanei\/\">\u03c0\u03bf\u03b9\u03bf component \u03c0\u03b1\u03c1\u03ae\u03b3\u03b1\u03b3\u03b5 \u03c4\u03bf \u03c0\u03c1\u03ce\u03c4\u03bf \u03b1\u03c0\u03bf\u03c6\u03b1\u03c3\u03b9\u03c3\u03c4\u03b9\u03ba\u03cc \u03bb\u03ac\u03b8\u03bf\u03c2<\/a>.<\/p>\n<h2 id=\"diakyvernisi-mnimis-paragogi\">\u0388\u03bd\u03b1 \u03bc\u03bf\u03bd\u03c4\u03ad\u03bb\u03bf \u03b4\u03b9\u03b1\u03ba\u03c5\u03b2\u03ad\u03c1\u03bd\u03b7\u03c3\u03b7\u03c2 \u03bc\u03bd\u03ae\u03bc\u03b7\u03c2 \u03b3\u03b9\u03b1 \u03c0\u03b1\u03c1\u03b1\u03b3\u03c9\u03b3\u03ae<\/h2>\n<p>\u0397 \u03bc\u03b5\u03bb\u03ad\u03c4\u03b7 \u03b4\u03b5\u03bd \u03b4\u03af\u03bd\u03b5\u03b9 \u03ad\u03c4\u03bf\u03b9\u03bc\u03bf \u03b5\u03c4\u03b1\u03b9\u03c1\u03b9\u03ba\u03cc framework, \u03b1\u03bb\u03bb\u03ac \u03c4\u03b1 \u03b5\u03c5\u03c1\u03ae\u03bc\u03b1\u03c4\u03b1 \u03c5\u03c0\u03bf\u03c3\u03c4\u03b7\u03c1\u03af\u03b6\u03bf\u03c5\u03bd \u03ba\u03b1\u03b8\u03b1\u03c1\u03ae \u03b1\u03c1\u03c7\u03b9\u03c4\u03b5\u03ba\u03c4\u03bf\u03bd\u03b9\u03ba\u03ae \u03b5\u03c5\u03b8\u03c5\u03bd\u03ce\u03bd. \u03a4\u03bf event store \u03ba\u03c1\u03b1\u03c4\u03ac \u03b1\u03bc\u03b5\u03c4\u03ac\u03b2\u03bb\u03b7\u03c4\u03b1 \u03c4\u03b1 raw \u03b5\u03c0\u03b5\u03b9\u03c3\u03cc\u03b4\u03b9\u03b1. \u03a4\u03bf consolidation layer \u03b4\u03b7\u03bc\u03b9\u03bf\u03c5\u03c1\u03b3\u03b5\u03af summaries \u03ae typed facts \u03c7\u03c9\u03c1\u03af\u03c2 \u03bd\u03b1 \u03b1\u03bd\u03c4\u03b9\u03ba\u03b1\u03b8\u03b9\u03c3\u03c4\u03ac \u03c4\u03b7\u03bd \u03c0\u03c1\u03c9\u03c4\u03bf\u03b3\u03b5\u03bd\u03ae \u03c0\u03b7\u03b3\u03ae. \u0388\u03bd\u03b1 policy layer \u03c7\u03b5\u03b9\u03c1\u03af\u03b6\u03b5\u03c4\u03b1\u03b9 scope, \u03b5\u03bc\u03c0\u03b9\u03c3\u03c4\u03bf\u03c3\u03cd\u03bd\u03b7, \u03bb\u03ae\u03be\u03b7 \u03ba\u03b1\u03b9 supersession. \u03a4\u03ad\u03bb\u03bf\u03c2, \u03c4\u03bf retrieval layer \u03b1\u03c0\u03bf\u03c6\u03b1\u03c3\u03af\u03b6\u03b5\u03b9 \u03c4\u03b9 \u03b5\u03c0\u03b9\u03c4\u03c1\u03ad\u03c0\u03b5\u03c4\u03b1\u03b9 \u03bd\u03b1 \u03bc\u03c0\u03b5\u03b9 \u03c3\u03c4\u03bf context \u03b3\u03b9\u03b1 \u03c4\u03b7 \u03c3\u03c5\u03b3\u03ba\u03b5\u03ba\u03c1\u03b9\u03bc\u03ad\u03bd\u03b7 \u03b5\u03c1\u03b3\u03b1\u03c3\u03af\u03b1.<\/p>\n<p>\u039a\u03ac\u03b8\u03b5 \u03bc\u03bd\u03ae\u03bc\u03b7 \u03c0\u03bf\u03c5 \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b5\u03bd\u03b5\u03c1\u03b3\u03bf\u03c0\u03bf\u03b9\u03ae\u03c3\u03b5\u03b9 write action \u03c7\u03c1\u03b5\u03b9\u03ac\u03b6\u03b5\u03c4\u03b1\u03b9 \u03b5\u03bb\u03ac\u03c7\u03b9\u03c3\u03c4\u03bf \u03c3\u03c5\u03bc\u03b2\u03cc\u03bb\u03b1\u03b9\u03bf: source event, timestamp, subject, scope, confidence, status \u03ba\u03b1\u03b9 \u03c3\u03cd\u03bd\u03b4\u03b5\u03c3\u03b7 \u03bc\u03b5 \u03c4\u03c5\u03c7\u03cc\u03bd superseding fact. \u0391\u03bd \u03bb\u03b5\u03af\u03c0\u03b5\u03b9 \u03c4\u03bf scope, \u03b7 \u03b1\u03c3\u03c6\u03b1\u03bb\u03ae\u03c2 \u03c3\u03c5\u03bc\u03c0\u03b5\u03c1\u03b9\u03c6\u03bf\u03c1\u03ac \u03b5\u03af\u03bd\u03b1\u03b9 \u03bd\u03b1 \u03bc\u03b7\u03bd \u03b5\u03c6\u03b1\u03c1\u03bc\u03bf\u03c3\u03c4\u03b5\u03af \u03bf \u03ba\u03b1\u03bd\u03cc\u03bd\u03b1\u03c2 \u03c3\u03b5 \u03b5\u03c5\u03c1\u03cd\u03c4\u03b5\u03c1\u03bf \u03c0\u03b5\u03b4\u03af\u03bf. \u0391\u03bd \u03b4\u03cd\u03bf facts \u03c3\u03c5\u03b3\u03ba\u03c1\u03bf\u03cd\u03bf\u03bd\u03c4\u03b1\u03b9, \u03c4\u03bf \u03c3\u03cd\u03c3\u03c4\u03b7\u03bc\u03b1 \u03c0\u03c1\u03ad\u03c0\u03b5\u03b9 \u03bd\u03b1 \u03b6\u03b7\u03c4\u03ae\u03c3\u03b5\u03b9 \u03b5\u03c0\u03b9\u03b2\u03b5\u03b2\u03b1\u03af\u03c9\u03c3\u03b7 \u03ae \u03bd\u03b1 \u03b1\u03ba\u03bf\u03bb\u03bf\u03c5\u03b8\u03ae\u03c3\u03b5\u03b9 \u03c3\u03b1\u03c6\u03ae \u03ba\u03b1\u03bd\u03cc\u03bd\u03b1 \u03c0\u03c1\u03bf\u03c4\u03b5\u03c1\u03b1\u03b9\u03cc\u03c4\u03b7\u03c4\u03b1\u03c2, \u03cc\u03c7\u03b9 \u03bd\u03b1 \u03b5\u03c0\u03b9\u03bb\u03ad\u03be\u03b5\u03b9 \u03c4\u03bf \u03c0\u03b9\u03bf \u03c0\u03b1\u03c1\u03cc\u03bc\u03bf\u03b9\u03bf embedding.<\/p>\n<p>\u03a7\u03c1\u03b5\u03b9\u03ac\u03b6\u03b5\u03c4\u03b1\u03b9 \u03b5\u03c0\u03af\u03c3\u03b7\u03c2 \u03c7\u03c9\u03c1\u03b9\u03c3\u03c4\u03cc audit log \u03b3\u03b9\u03b1 \u03c4\u03b9\u03c2 \u03b1\u03bd\u03b1\u03b3\u03bd\u03ce\u03c3\u03b5\u03b9\u03c2 \u03bc\u03bd\u03ae\u03bc\u03b7\u03c2. \u0394\u03b5\u03bd \u03b1\u03c1\u03ba\u03b5\u03af \u03bd\u03b1 \u03b3\u03bd\u03c9\u03c1\u03af\u03b6\u03b5\u03b9 \u03b7 \u03bf\u03bc\u03ac\u03b4\u03b1 \u03c4\u03b9 \u03b1\u03c0\u03bf\u03b8\u03b7\u03ba\u03b5\u03cd\u03c4\u03b7\u03ba\u03b5\u00b7 \u03c0\u03c1\u03ad\u03c0\u03b5\u03b9 \u03bd\u03b1 \u03b2\u03bb\u03ad\u03c0\u03b5\u03b9 \u03c4\u03b9 \u03b1\u03bd\u03b1\u03ba\u03c4\u03ae\u03b8\u03b7\u03ba\u03b5 \u03c0\u03c1\u03b9\u03bd \u03b1\u03c0\u03cc \u03bc\u03b9\u03b1 \u03b1\u03bb\u03bb\u03b1\u03b3\u03ae, \u03c4\u03b9 \u03b1\u03c0\u03bf\u03c1\u03c1\u03af\u03c6\u03b8\u03b7\u03ba\u03b5 \u03b1\u03c0\u03cc \u03c4\u03bf gate \u03ba\u03b1\u03b9 \u03c0\u03bf\u03b9\u03b1 \u03c0\u03b7\u03b3\u03ae \u03b5\u03c0\u03b7\u03c1\u03ad\u03b1\u03c3\u03b5 \u03c4\u03bf \u03c4\u03b5\u03bb\u03b9\u03ba\u03cc action. \u0391\u03c5\u03c4\u03ae \u03b7 \u03bf\u03c1\u03b1\u03c4\u03cc\u03c4\u03b7\u03c4\u03b1 \u03be\u03b5\u03c7\u03c9\u03c1\u03af\u03b6\u03b5\u03b9 \u03b1\u03c3\u03c4\u03bf\u03c7\u03af\u03b1 \u03bc\u03bf\u03bd\u03c4\u03ad\u03bb\u03bf\u03c5 \u03b1\u03c0\u03cc \u03b1\u03c3\u03c4\u03bf\u03c7\u03af\u03b1 retrieval \u03ae \u03bb\u03ac\u03b8\u03bf\u03c2 consolidation.<\/p>\n<div class=\"td-decision-band\">\n<p class=\"td-decision-label\">Gate \u03c0\u03c1\u03b9\u03bd \u03b7 \u03bc\u03bd\u03ae\u03bc\u03b7 \u03b5\u03bd\u03b5\u03c1\u03b3\u03bf\u03c0\u03bf\u03b9\u03ae\u03c3\u03b5\u03b9 write action<\/p>\n<p><strong>\u039c\u03b9\u03b1 \u03b1\u03bd\u03ac\u03bc\u03bd\u03b7\u03c3\u03b7 \u03c7\u03c9\u03c1\u03af\u03c2 provenance, scope \u03ae status \u03b4\u03b5\u03bd \u03b5\u03af\u03bd\u03b1\u03b9 \u03ad\u03b3\u03ba\u03c5\u03c1\u03bf\u03c2 \u03b5\u03c0\u03b9\u03c7\u03b5\u03b9\u03c1\u03b7\u03c3\u03b9\u03b1\u03ba\u03cc\u03c2 \u03ba\u03b1\u03bd\u03cc\u03bd\u03b1\u03c2.<\/strong><\/p>\n<p>\u039c\u03c0\u03bb\u03bf\u03ba\u03ac\u03c1\u03b5\u03c4\u03b5 \u03c4\u03b7\u03bd \u03b1\u03c5\u03c4\u03bf\u03bd\u03bf\u03bc\u03af\u03b1 \u03cc\u03c4\u03b1\u03bd \u03b7 \u03c0\u03b7\u03b3\u03ae \u03b4\u03b5\u03bd \u03b5\u03af\u03bd\u03b1\u03b9 \u03b5\u03bb\u03ad\u03b3\u03be\u03b9\u03bc\u03b7, \u03cc\u03c4\u03b1\u03bd \u03c5\u03c0\u03ac\u03c1\u03c7\u03b5\u03b9 \u03bd\u03b5\u03cc\u03c4\u03b5\u03c1\u03bf conflicting fact \u03ae \u03cc\u03c4\u03b1\u03bd \u03c4\u03bf retrieval gate \u03b4\u03b5\u03bd \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b1\u03b9\u03c4\u03b9\u03bf\u03bb\u03bf\u03b3\u03ae\u03c3\u03b5\u03b9 \u03b3\u03b9\u03b1\u03c4\u03af \u03b7 \u03bc\u03bd\u03ae\u03bc\u03b7 \u03b1\u03c6\u03bf\u03c1\u03ac \u03c4\u03b7 \u03c3\u03c5\u03b3\u03ba\u03b5\u03ba\u03c1\u03b9\u03bc\u03ad\u03bd\u03b7 \u03b5\u03c1\u03b3\u03b1\u03c3\u03af\u03b1. \u0395\u03c0\u03b9\u03c4\u03c1\u03ad\u03c8\u03c4\u03b5 action \u03bc\u03cc\u03bd\u03bf \u03bc\u03b1\u03b6\u03af \u03bc\u03b5 \u03b1\u03bd\u03b5\u03be\u03ac\u03c1\u03c4\u03b7\u03c4\u03bf verifier \u03ba\u03b1\u03b9 \u03b1\u03bd\u03b1\u03c3\u03c4\u03c1\u03ad\u03c8\u03b9\u03bc\u03b7 \u03b4\u03b9\u03b1\u03b4\u03c1\u03bf\u03bc\u03ae.<\/p>\n<\/div>\n<p>\u03a4\u03b1 \u03b4\u03b9\u03ba\u03b1\u03b9\u03ce\u03bc\u03b1\u03c4\u03b1 \u03c0\u03c1\u03ad\u03c0\u03b5\u03b9 \u03bd\u03b1 \u03b1\u03ba\u03bf\u03bb\u03bf\u03c5\u03b8\u03bf\u03cd\u03bd \u03c4\u03bf \u03af\u03b4\u03b9\u03bf scope. \u039c\u03bd\u03ae\u03bc\u03b7 \u03c0\u03bf\u03c5 \u03c0\u03c1\u03bf\u03ad\u03c1\u03c7\u03b5\u03c4\u03b1\u03b9 \u03b1\u03c0\u03cc customer-support ticket \u03b4\u03b5\u03bd \u03c0\u03c1\u03ad\u03c0\u03b5\u03b9 \u03bd\u03b1 \u03b1\u03c0\u03bf\u03ba\u03c4\u03ac \u03b5\u03be\u03bf\u03c5\u03c3\u03af\u03b1 \u03c0\u03ac\u03bd\u03c9 \u03c3\u03b5 global CRM policy. \u0397 \u03b1\u03c1\u03c7\u03b9\u03c4\u03b5\u03ba\u03c4\u03bf\u03bd\u03b9\u03ba\u03ae \u03c4\u03c9\u03bd <a href=\"https:\/\/twodots.gr\/dynamika-dikaiomata-ai-agents-elachisti-prosvasi\/\">\u03b4\u03c5\u03bd\u03b1\u03bc\u03b9\u03ba\u03ce\u03bd \u03b4\u03b9\u03ba\u03b1\u03b9\u03c9\u03bc\u03ac\u03c4\u03c9\u03bd \u03b5\u03bb\u03ac\u03c7\u03b9\u03c3\u03c4\u03b7\u03c2 \u03c0\u03c1\u03cc\u03c3\u03b2\u03b1\u03c3\u03b7\u03c2<\/a> \u03c0\u03b5\u03c1\u03b9\u03bf\u03c1\u03af\u03b6\u03b5\u03b9 credentials, \u03b4\u03b9\u03ac\u03c1\u03ba\u03b5\u03b9\u03b1 \u03ba\u03b1\u03b9 blast radius, \u03b5\u03bd\u03ce \u03c4\u03bf <a href=\"https:\/\/twodots.gr\/agentic-ai-security-stack-controls-technologikes-ependyseis\/\">agentic AI security stack<\/a> \u03c3\u03c5\u03bd\u03b4\u03ad\u03b5\u03b9 identity, data, runtime \u03ba\u03b1\u03b9 monitoring controls.<\/p>\n<h2 id=\"pilot-axiologisis-mnimis\">\u03a0\u03ce\u03c2 \u03c3\u03c4\u03ae\u03bd\u03b5\u03c4\u03b1\u03b9 pilot \u03b1\u03be\u03b9\u03bf\u03bb\u03cc\u03b3\u03b7\u03c3\u03b7\u03c2 \u03bc\u03bd\u03ae\u03bc\u03b7\u03c2<\/h2>\n<p>\u0397 \u03c5\u03b9\u03bf\u03b8\u03ad\u03c4\u03b7\u03c3\u03b7 \u03b5\u03be\u03c9\u03c4\u03b5\u03c1\u03b9\u03ba\u03ae\u03c2 \u03bc\u03bd\u03ae\u03bc\u03b7\u03c2 \u03b4\u03b5\u03bd \u03c0\u03c1\u03ad\u03c0\u03b5\u03b9 \u03bd\u03b1 \u03be\u03b5\u03ba\u03b9\u03bd\u03ac \u03b1\u03c0\u03cc \u03c4\u03bf \u03c0\u03cc\u03c3\u03b1 tokens \u03ae vectors \u03c7\u03c9\u03c1\u03ac. \u039e\u03b5\u03ba\u03b9\u03bd\u03ac \u03b1\u03c0\u03cc \u03c4\u03b9\u03c2 \u03b1\u03c0\u03bf\u03c4\u03c5\u03c7\u03af\u03b5\u03c2 \u03c0\u03bf\u03c5 \u03b8\u03ad\u03bb\u03b5\u03b9 \u03bd\u03b1 \u03b1\u03c0\u03bf\u03c4\u03c1\u03ad\u03c8\u03b5\u03b9 \u03b7 \u03bf\u03bc\u03ac\u03b4\u03b1. \u03a7\u03c1\u03b5\u03b9\u03ac\u03b6\u03bf\u03bd\u03c4\u03b1\u03b9 \u03c3\u03b5\u03bd\u03ac\u03c1\u03b9\u03b1 \u03cc\u03c0\u03bf\u03c5 \u03bd\u03b5\u03cc\u03c4\u03b5\u03c1\u03b7 \u03bf\u03b4\u03b7\u03b3\u03af\u03b1 \u03b1\u03ba\u03c5\u03c1\u03ce\u03bd\u03b5\u03b9 \u03c0\u03b1\u03bb\u03b9\u03cc\u03c4\u03b5\u03c1\u03b7, \u03ad\u03bd\u03b1 feedback \u03b9\u03c3\u03c7\u03cd\u03b5\u03b9 \u03bc\u03cc\u03bd\u03bf \u03c3\u03b5 \u03c0\u03b5\u03c1\u03b9\u03bf\u03c1\u03b9\u03c3\u03bc\u03ad\u03bd\u03bf scope \u03ba\u03b1\u03b9 \u03b7 \u03c3\u03c9\u03c3\u03c4\u03ae \u03b5\u03bd\u03ad\u03c1\u03b3\u03b5\u03b9\u03b1 \u03b1\u03c0\u03b1\u03b9\u03c4\u03b5\u03af \u03b1\u03ba\u03c1\u03b9\u03b2\u03ad\u03c2 \u03c3\u03c4\u03bf\u03b9\u03c7\u03b5\u03af\u03bf \u03b1\u03c0\u03cc \u03c0\u03c1\u03bf\u03b7\u03b3\u03bf\u03cd\u03bc\u03b5\u03bd\u03b7 \u03c3\u03c5\u03bd\u03b5\u03b4\u03c1\u03af\u03b1.<\/p>\n<div class=\"td-step-list\">\n<p class=\"td-step-list-title\">\u0395\u03c0\u03c4\u03ac \u03b2\u03ae\u03bc\u03b1\u03c4\u03b1 \u03b3\u03b9\u03b1 pilot \u03bc\u03bd\u03ae\u03bc\u03b7\u03c2 AI agent<\/p>\n<ol>\n<li><span class=\"td-step-kicker\">Step 1<\/span><strong>\u039f\u03c1\u03af\u03c3\u03c4\u03b5 \u03c4\u03bf action \u03ba\u03b1\u03b9 \u03c4\u03bf blast radius<\/strong>\n<p>\u0394\u03b9\u03b1\u03bb\u03ad\u03be\u03c4\u03b5 \u03ad\u03bd\u03b1 \u03c3\u03c5\u03b3\u03ba\u03b5\u03ba\u03c1\u03b9\u03bc\u03ad\u03bd\u03bf workflow, \u03c4\u03b1 \u03c3\u03c5\u03c3\u03c4\u03ae\u03bc\u03b1\u03c4\u03b1 \u03c0\u03bf\u03c5 \u03b4\u03b9\u03b1\u03b2\u03ac\u03b6\u03b5\u03b9 \u03ae \u03b3\u03c1\u03ac\u03c6\u03b5\u03b9 \u03bf agent \u03ba\u03b1\u03b9 \u03c4\u03b9\u03c2 \u03c3\u03c5\u03bd\u03ad\u03c0\u03b5\u03b9\u03b5\u03c2 \u03bc\u03b9\u03b1\u03c2 stale \u03ae overscoped \u03bc\u03bd\u03ae\u03bc\u03b7\u03c2.<\/p>\n<\/li>\n<li><span class=\"td-step-kicker\">Step 2<\/span><strong>\u0394\u03b7\u03bc\u03b9\u03bf\u03c5\u03c1\u03b3\u03ae\u03c3\u03c4\u03b5 traps \u03b1\u03c0\u03cc \u03c0\u03c1\u03b1\u03b3\u03bc\u03b1\u03c4\u03b9\u03ba\u03ad\u03c2 \u03b1\u03c0\u03bf\u03c4\u03c5\u03c7\u03af\u03b5\u03c2<\/strong>\n<p>\u03a0\u03c1\u03bf\u03c3\u03b8\u03ad\u03c3\u03c4\u03b5 superseded \u03bf\u03b4\u03b7\u03b3\u03af\u03b5\u03c2, scoped feedback, generated-file boundaries, irrelevant context \u03ba\u03b1\u03b9 \u03c0\u03b5\u03c1\u03b9\u03c0\u03c4\u03ce\u03c3\u03b5\u03b9\u03c2 \u03cc\u03c0\u03bf\u03c5 \u03b7 \u03c3\u03c9\u03c3\u03c4\u03ae \u03b5\u03bd\u03ad\u03c1\u03b3\u03b5\u03b9\u03b1 \u03b5\u03af\u03bd\u03b1\u03b9 \u03b1\u03c0\u03bf\u03c7\u03ae.<\/p>\n<\/li>\n<li><span class=\"td-step-kicker\">Step 3<\/span><strong>\u03a0\u03b1\u03b3\u03ce\u03c3\u03c4\u03b5 \u03c4\u03b1 executable outcomes<\/strong>\n<p>\u039f\u03c1\u03af\u03c3\u03c4\u03b5 tests, policy checks \u03ae deterministic validations \u03c0\u03c1\u03b9\u03bd \u03b1\u03c0\u03cc \u03c4\u03b1 runs, \u03ce\u03c3\u03c4\u03b5 \u03b7 \u03b5\u03c0\u03b9\u03c4\u03c5\u03c7\u03af\u03b1 \u03bd\u03b1 \u03bc\u03b7 \u03b2\u03b1\u03b8\u03bc\u03bf\u03bb\u03bf\u03b3\u03b5\u03af\u03c4\u03b1\u03b9 \u03b1\u03c0\u03cc \u03c0\u03b5\u03b9\u03c3\u03c4\u03b9\u03ba\u03ae \u03b3\u03bb\u03ce\u03c3\u03c3\u03b1.<\/p>\n<\/li>\n<li><span class=\"td-step-kicker\">Step 4<\/span><strong>\u039a\u03b1\u03c4\u03b1\u03b3\u03c1\u03ac\u03c8\u03c4\u03b5 provenance \u03ba\u03b1\u03b9 supersession<\/strong>\n<p>\u03a3\u03c5\u03bd\u03b4\u03ad\u03c3\u03c4\u03b5 \u03ba\u03ac\u03b8\u03b5 derived memory \u03bc\u03b5 \u03c4\u03bf raw event, \u03c7\u03c1\u03cc\u03bd\u03bf, subject, scope, confidence, status \u03ba\u03b1\u03b9 \u03c4\u03c5\u03c7\u03cc\u03bd \u03bd\u03b5\u03cc\u03c4\u03b5\u03c1\u03bf fact \u03c0\u03bf\u03c5 \u03c4\u03b7\u03bd \u03b1\u03ba\u03c5\u03c1\u03ce\u03bd\u03b5\u03b9.<\/p>\n<\/li>\n<li><span class=\"td-step-kicker\">Step 5<\/span><strong>\u039c\u03b5\u03c4\u03c1\u03ae\u03c3\u03c4\u03b5 harmful retrieval \u03c7\u03c9\u03c1\u03b9\u03c3\u03c4\u03ac<\/strong>\n<p>\u039a\u03b1\u03c4\u03b1\u03b3\u03c1\u03ac\u03c8\u03c4\u03b5 stale use, overscope, irrelevant import \u03ba\u03b1\u03b9 repeated error \u03b1\u03bd\u03c4\u03af \u03bd\u03b1 \u03ba\u03c1\u03cd\u03b2\u03bf\u03bd\u03c4\u03b1\u03b9 \u03bc\u03ad\u03c3\u03b1 \u03c3\u03b5 \u03ad\u03bd\u03b1 \u03c3\u03c5\u03bd\u03bf\u03bb\u03b9\u03ba\u03cc pass rate.<\/p>\n<\/li>\n<li><span class=\"td-step-kicker\">Step 6<\/span><strong>\u0395\u03bb\u03ad\u03b3\u03be\u03c4\u03b5 retrieval \u03ba\u03b1\u03b9 action trace<\/strong>\n<p>\u039a\u03c1\u03b1\u03c4\u03ae\u03c3\u03c4\u03b5 \u03c4\u03b9 \u03b1\u03bd\u03b1\u03ba\u03c4\u03ae\u03b8\u03b7\u03ba\u03b5, \u03c4\u03b9 \u03bc\u03c0\u03bb\u03bf\u03ba\u03b1\u03c1\u03af\u03c3\u03c4\u03b7\u03ba\u03b5, \u03c0\u03bf\u03b9\u03b1 \u03c0\u03b7\u03b3\u03ae \u03b5\u03c0\u03b7\u03c1\u03ad\u03b1\u03c3\u03b5 \u03c4\u03b7\u03bd \u03b1\u03c0\u03cc\u03c6\u03b1\u03c3\u03b7, \u03c0\u03bf\u03b9\u03bf write \u03ad\u03b3\u03b9\u03bd\u03b5 \u03ba\u03b1\u03b9 \u03c0\u03bf\u03b9\u03bf oracle \u03c4\u03bf \u03b5\u03c0\u03b1\u03bb\u03ae\u03b8\u03b5\u03c5\u03c3\u03b5.<\/p>\n<\/li>\n<li><span class=\"td-step-kicker\">Step 7<\/span><strong>\u0391\u03c0\u03bf\u03c6\u03b1\u03c3\u03af\u03c3\u03c4\u03b5 expand, constrain \u03ae stop<\/strong>\n<p>\u0391\u03c5\u03be\u03ae\u03c3\u03c4\u03b5 autonomy \u03bc\u03cc\u03bd\u03bf \u03b1\u03bd task success, harmful retrieval, latency, \u03ba\u03cc\u03c3\u03c4\u03bf\u03c2, auditability \u03ba\u03b1\u03b9 rollback \u03c0\u03b1\u03c1\u03b1\u03bc\u03ad\u03bd\u03bf\u03c5\u03bd \u03b1\u03c0\u03bf\u03b4\u03b5\u03ba\u03c4\u03ac \u03c3\u03b5 \u03b5\u03c0\u03b1\u03bd\u03b1\u03bb\u03b1\u03bc\u03b2\u03b1\u03bd\u03cc\u03bc\u03b5\u03bd\u03b1 runs.<\/p>\n<\/li>\n<\/ol>\n<\/div>\n<p>\u0391\u03c5\u03c4\u03ae \u03b7 \u03b4\u03b9\u03b1\u03b4\u03b9\u03ba\u03b1\u03c3\u03af\u03b1 \u03bc\u03b5\u03c4\u03b1\u03c4\u03c1\u03ad\u03c0\u03b5\u03b9 \u03c4\u03b7 \u03bc\u03bd\u03ae\u03bc\u03b7 \u03b1\u03c0\u03cc feature demo \u03c3\u03b5 \u03b5\u03bb\u03b5\u03b3\u03c7\u03cc\u03bc\u03b5\u03bd\u03bf \u03c5\u03c0\u03bf\u03c3\u03cd\u03c3\u03c4\u03b7\u03bc\u03b1. \u0395\u03b9\u03b4\u03b9\u03ba\u03ac \u03cc\u03c4\u03b1\u03bd \u03bf agent \u03b3\u03c1\u03ac\u03c6\u03b5\u03b9 \u03ba\u03ce\u03b4\u03b9\u03ba\u03b1, \u03b5\u03bd\u03b7\u03bc\u03b5\u03c1\u03ce\u03bd\u03b5\u03b9 CRM \u03ae \u03b1\u03bb\u03bb\u03ac\u03b6\u03b5\u03b9 \u03c0\u03b5\u03c1\u03b9\u03b5\u03c7\u03cc\u03bc\u03b5\u03bd\u03bf, \u03b7 \u03c0\u03bf\u03b9\u03cc\u03c4\u03b7\u03c4\u03b1 \u03c4\u03b7\u03c2 \u03b1\u03bd\u03ac\u03ba\u03c4\u03b7\u03c3\u03b7\u03c2 \u03c0\u03c1\u03ad\u03c0\u03b5\u03b9 \u03bd\u03b1 \u03c3\u03c5\u03bd\u03b4\u03ad\u03b5\u03c4\u03b1\u03b9 \u03bc\u03b5 \u03c3\u03b1\u03c6\u03ae \u03cc\u03c1\u03b9\u03b1 \u03ba\u03b1\u03b9 \u03b4\u03c5\u03bd\u03b1\u03c4\u03cc\u03c4\u03b7\u03c4\u03b1 \u03b1\u03bd\u03b1\u03b4\u03c1\u03bf\u03bc\u03b9\u03ba\u03bf\u03cd \u03b5\u03bb\u03ad\u03b3\u03c7\u03bf\u03c5. \u0397 \u03b1\u03be\u03b9\u03bf\u03bb\u03cc\u03b3\u03b7\u03c3\u03b7 \u03b5\u03bd\u03cc\u03c2 <a href=\"https:\/\/twodots.gr\/terminal-agents-ai-command-line-plires-systima\/\">terminal agent \u03c9\u03c2 \u03c0\u03bb\u03ae\u03c1\u03bf\u03c5\u03c2 \u03c3\u03c5\u03c3\u03c4\u03ae\u03bc\u03b1\u03c4\u03bf\u03c2<\/a> \u03c0\u03c1\u03bf\u03c3\u03c6\u03ad\u03c1\u03b5\u03b9 \u03c7\u03c1\u03ae\u03c3\u03b9\u03bc\u03bf \u03c0\u03c1\u03cc\u03c4\u03c5\u03c0\u03bf: model, harness, runtime, verifier \u03ba\u03b1\u03b9 permissions \u03b1\u03bd\u03b1\u03c6\u03ad\u03c1\u03bf\u03bd\u03c4\u03b1\u03b9 \u03bc\u03b1\u03b6\u03af.<\/p>\n<h2 id=\"kostos-asfaleia\">\u039a\u03cc\u03c3\u03c4\u03bf\u03c2, \u03b1\u03c3\u03c6\u03ac\u03bb\u03b5\u03b9\u03b1 \u03ba\u03b1\u03b9 \u03b5\u03c0\u03b9\u03c7\u03b5\u03b9\u03c1\u03b7\u03c3\u03b9\u03b1\u03ba\u03cc \u03b1\u03c0\u03bf\u03c4\u03ad\u03bb\u03b5\u03c3\u03bc\u03b1<\/h2>\n<p>\u0397 \u03bc\u03b5\u03bb\u03ad\u03c4\u03b7 \u03c0\u03b5\u03c1\u03b9\u03bb\u03b1\u03bc\u03b2\u03ac\u03bd\u03b5\u03b9 \u03b4\u03b9\u03b1\u03b3\u03bd\u03c9\u03c3\u03c4\u03b9\u03ba\u03ac \u03ba\u03cc\u03c3\u03c4\u03bf\u03c5\u03c2. \u03a3\u03c4\u03bf v2 \u03b1\u03bd\u03b1\u03c6\u03ad\u03c1\u03b5\u03b9 \u03ba\u03cc\u03c3\u03c4\u03bf\u03c2 \u03b1\u03bd\u03ac \u03b5\u03c0\u03b9\u03c4\u03c5\u03c7\u03b7\u03bc\u03ad\u03bd\u03b7 \u03b5\u03c1\u03b3\u03b1\u03c3\u03af\u03b1 0,0356 \u03b4\u03bf\u03bb\u03ac\u03c1\u03b9\u03b1 \u03b3\u03b9\u03b1 raw-only reference probe, 0,0383 \u03b3\u03b9\u03b1 hybrid \u03ba\u03b1\u03b9 0,0395 \u03b3\u03b9\u03b1 B5. \u039f\u03b9 \u03c4\u03b9\u03bc\u03ad\u03c2 \u03b1\u03c6\u03bf\u03c1\u03bf\u03cd\u03bd \u03c4\u03bf \u03c3\u03c5\u03b3\u03ba\u03b5\u03ba\u03c1\u03b9\u03bc\u03ad\u03bd\u03bf \u03c0\u03b5\u03af\u03c1\u03b1\u03bc\u03b1, \u03c4\u03b9\u03c2 \u03c3\u03c5\u03b3\u03ba\u03b5\u03ba\u03c1\u03b9\u03bc\u03ad\u03bd\u03b5\u03c2 \u03c5\u03bb\u03bf\u03c0\u03bf\u03b9\u03ae\u03c3\u03b5\u03b9\u03c2 \u03ba\u03b1\u03b9 \u03c4\u03b7\u03bd \u03c0\u03b5\u03c1\u03af\u03bf\u03b4\u03bf \u03b5\u03ba\u03c4\u03ad\u03bb\u03b5\u03c3\u03b7\u03c2. \u0394\u03b5\u03bd \u03b5\u03af\u03bd\u03b1\u03b9 \u03c4\u03c1\u03ad\u03c7\u03c9\u03bd \u03c4\u03b9\u03bc\u03bf\u03ba\u03b1\u03c4\u03ac\u03bb\u03bf\u03b3\u03bf\u03c2 \u03ae \u03c0\u03c1\u03cc\u03b2\u03bb\u03b5\u03c8\u03b7 \u03c0\u03b1\u03c1\u03b1\u03b3\u03c9\u03b3\u03b9\u03ba\u03bf\u03cd \u03ba\u03cc\u03c3\u03c4\u03bf\u03c5\u03c2.<\/p>\n<p>\u03a4\u03bf \u03c7\u03c1\u03ae\u03c3\u03b9\u03bc\u03bf \u03bc\u03ac\u03b8\u03b7\u03bc\u03b1 \u03b5\u03af\u03bd\u03b1\u03b9 \u03bc\u03b5\u03b8\u03bf\u03b4\u03bf\u03bb\u03bf\u03b3\u03b9\u03ba\u03cc: \u03c4\u03bf \u03ba\u03cc\u03c3\u03c4\u03bf\u03c2 \u03c0\u03c1\u03ad\u03c0\u03b5\u03b9 \u03bd\u03b1 \u03c3\u03c5\u03c3\u03c7\u03b5\u03c4\u03af\u03b6\u03b5\u03c4\u03b1\u03b9 \u03bc\u03b5 \u03b5\u03c0\u03b9\u03c4\u03c5\u03c7\u03b5\u03af\u03c2 \u03b5\u03c1\u03b3\u03b1\u03c3\u03af\u03b5\u03c2, \u03cc\u03c7\u03b9 \u03bc\u03cc\u03bd\u03bf \u03bc\u03b5 \u03c4\u03bf \u03ba\u03cc\u03c3\u03c4\u03bf\u03c2 \u03ba\u03ac\u03b8\u03b5 \u03ba\u03bb\u03ae\u03c3\u03b7\u03c2. \u039c\u03b9\u03b1 \u03c6\u03b8\u03b7\u03bd\u03cc\u03c4\u03b5\u03c1\u03b7 \u03bb\u03cd\u03c3\u03b7 \u03c0\u03bf\u03c5 \u03b5\u03b9\u03c3\u03ac\u03b3\u03b5\u03b9 stale \u03bf\u03b4\u03b7\u03b3\u03af\u03b5\u03c2 \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b4\u03b7\u03bc\u03b9\u03bf\u03c5\u03c1\u03b3\u03ae\u03c3\u03b5\u03b9 downstream \u03ba\u03cc\u03c3\u03c4\u03bf\u03c2 \u03c3\u03b5 review, rollback \u03ba\u03b1\u03b9 \u03b4\u03b9\u03cc\u03c1\u03b8\u03c9\u03c3\u03b7. \u0393\u03b9\u03b1 \u03c3\u03c9\u03c3\u03c4\u03ae \u03b1\u03c0\u03cc\u03c6\u03b1\u03c3\u03b7 \u03c7\u03c1\u03b5\u03b9\u03ac\u03b6\u03bf\u03bd\u03c4\u03b1\u03b9 \u03bc\u03b1\u03b6\u03af success rate, harmful-retrieval metrics, latency, auditability \u03ba\u03b1\u03b9 \u03bb\u03b5\u03b9\u03c4\u03bf\u03c5\u03c1\u03b3\u03b9\u03ba\u03cc \u03ba\u03cc\u03c3\u03c4\u03bf\u03c2.<\/p>\n<p>\u0397 \u03bc\u03bd\u03ae\u03bc\u03b7 \u03b5\u03af\u03bd\u03b1\u03b9 \u03ba\u03b1\u03b9 \u03b5\u03c0\u03b9\u03c6\u03ac\u03bd\u03b5\u03b9\u03b1 \u03b5\u03c0\u03af\u03b8\u03b5\u03c3\u03b7\u03c2. \u03a4\u03bf OWASP ASI06 \u03c0\u03b5\u03c1\u03b9\u03b3\u03c1\u03ac\u03c6\u03b5\u03b9 memory and context poisoning: \u03bc\u03b7 \u03b1\u03be\u03b9\u03cc\u03c0\u03b9\u03c3\u03c4\u03bf \u03c0\u03b5\u03c1\u03b9\u03b5\u03c7\u03cc\u03bc\u03b5\u03bd\u03bf \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03c0\u03b1\u03c1\u03b1\u03bc\u03b5\u03af\u03bd\u03b5\u03b9, \u03bd\u03b1 \u03b1\u03bd\u03b1\u03ba\u03c4\u03b7\u03b8\u03b5\u03af \u03b1\u03c1\u03b3\u03cc\u03c4\u03b5\u03c1\u03b1 \u03ba\u03b1\u03b9 \u03bd\u03b1 \u03b5\u03c0\u03b7\u03c1\u03b5\u03ac\u03c3\u03b5\u03b9 planning \u03ae tool use. \u0393\u03b9\u2019 \u03b1\u03c5\u03c4\u03cc ingestion validation, separation, provenance, read logs \u03ba\u03b1\u03b9 \u03b4\u03c5\u03bd\u03b1\u03c4\u03cc\u03c4\u03b7\u03c4\u03b1 \u03b1\u03bd\u03ac\u03ba\u03bb\u03b7\u03c3\u03b7\u03c2 \u03b4\u03b5\u03bd \u03b5\u03af\u03bd\u03b1\u03b9 \u03bc\u03cc\u03bd\u03bf \u03c0\u03bf\u03b9\u03bf\u03c4\u03b9\u03ba\u03ac \u03c7\u03b1\u03c1\u03b1\u03ba\u03c4\u03b7\u03c1\u03b9\u03c3\u03c4\u03b9\u03ba\u03ac\u00b7 \u03b5\u03af\u03bd\u03b1\u03b9 security controls.<\/p>\n<p>\u03a4\u03bf NIST AI RMF Generative AI Profile \u03c4\u03bf\u03c0\u03bf\u03b8\u03b5\u03c4\u03b5\u03af testing, evaluation, verification \u03ba\u03b1\u03b9 validation \u03bc\u03ad\u03c3\u03b1 \u03c3\u03b5 \u03c3\u03c5\u03bd\u03b5\u03c7\u03ae \u03b4\u03b9\u03b1\u03c7\u03b5\u03af\u03c1\u03b9\u03c3\u03b7 \u03c1\u03af\u03c3\u03ba\u03bf\u03c5. \u0393\u03b9\u03b1 \u03ad\u03bd\u03b1\u03bd memory-bearing agent, \u03b1\u03c5\u03c4\u03cc \u03c3\u03b7\u03bc\u03b1\u03af\u03bd\u03b5\u03b9 \u03cc\u03c4\u03b9 \u03b7 \u03bf\u03bc\u03ac\u03b4\u03b1 \u03c0\u03c1\u03ad\u03c0\u03b5\u03b9 \u03bd\u03b1 \u03b5\u03bb\u03ad\u03b3\u03c7\u03b5\u03b9 \u03c4\u03bf \u03c0\u03bb\u03ae\u03c1\u03b5\u03c2 lifecycle \u03c4\u03b7\u03c2 \u03bc\u03bd\u03ae\u03bc\u03b7\u03c2: \u03b5\u03b9\u03c3\u03b1\u03b3\u03c9\u03b3\u03ae, consolidation, \u03b1\u03bd\u03ac\u03ba\u03c4\u03b7\u03c3\u03b7, \u03c7\u03c1\u03ae\u03c3\u03b7, \u03bb\u03ae\u03be\u03b7, \u03b4\u03b9\u03cc\u03c1\u03b8\u03c9\u03c3\u03b7 \u03ba\u03b1\u03b9 \u03b4\u03b9\u03b1\u03b3\u03c1\u03b1\u03c6\u03ae \u03cc\u03c0\u03bf\u03c5 \u03b1\u03c0\u03b1\u03b9\u03c4\u03b5\u03af\u03c4\u03b1\u03b9 \u03b1\u03c0\u03cc \u03c4\u03b7\u03bd \u03c0\u03bf\u03bb\u03b9\u03c4\u03b9\u03ba\u03ae \u03b4\u03b5\u03b4\u03bf\u03bc\u03ad\u03bd\u03c9\u03bd.<\/p>\n<h2 id=\"oria-kai-symperasma\">\u03a4\u03b1 \u03cc\u03c1\u03b9\u03b1 \u03ba\u03b1\u03b9 \u03c4\u03bf \u03c3\u03c4\u03c1\u03b1\u03c4\u03b7\u03b3\u03b9\u03ba\u03cc \u03c3\u03c5\u03bc\u03c0\u03ad\u03c1\u03b1\u03c3\u03bc\u03b1<\/h2>\n<p>\u0397 \u03b5\u03c1\u03b3\u03b1\u03c3\u03af\u03b1 \u03c7\u03c1\u03b7\u03c3\u03b9\u03bc\u03bf\u03c0\u03bf\u03b9\u03b5\u03af \u03ad\u03bd\u03b1\u03bd wake model \u03ba\u03b1\u03b9 \u03b4\u03b5\u03bd \u03b1\u03c0\u03bf\u03b4\u03b5\u03b9\u03ba\u03bd\u03cd\u03b5\u03b9 \u03bc\u03b5\u03c4\u03b1\u03c6\u03bf\u03c1\u03ac \u03c4\u03c9\u03bd \u03b1\u03c0\u03bf\u03c4\u03b5\u03bb\u03b5\u03c3\u03bc\u03ac\u03c4\u03c9\u03bd \u03c3\u03b5 \u03ac\u03bb\u03bb\u03b1 \u03bc\u03bf\u03bd\u03c4\u03ad\u03bb\u03b1. \u03a4\u03bf benchmark \u03b5\u03af\u03bd\u03b1\u03b9 \u03b5\u03bb\u03b5\u03b3\u03c7\u03cc\u03bc\u03b5\u03bd\u03bf \u03ba\u03b1\u03b9 \u03b5\u03c3\u03c4\u03b9\u03ac\u03b6\u03b5\u03b9 \u03c3\u03b5 \u03c3\u03c5\u03b3\u03ba\u03b5\u03ba\u03c1\u03b9\u03bc\u03ad\u03bd\u03b5\u03c2 \u03c0\u03b1\u03b8\u03bf\u03bb\u03bf\u03b3\u03af\u03b5\u03c2 software engineering. \u0397 \u03b1\u03c0\u03bf\u03bc\u03cc\u03bd\u03c9\u03c3\u03b7 \u03b1\u03c6\u03bf\u03c1\u03ac \u03c4\u03bf filesystem, \u03cc\u03c7\u03b9 \u03c0\u03bb\u03ae\u03c1\u03b7 network isolation. \u039f\u03b9 baseline \u03c5\u03bb\u03bf\u03c0\u03bf\u03b9\u03ae\u03c3\u03b5\u03b9\u03c2 \u03b5\u03af\u03bd\u03b1\u03b9 \u03bf\u03c1\u03b9\u03bf\u03b8\u03b5\u03c4\u03b7\u03bc\u03ad\u03bd\u03b5\u03c2 \u03ba\u03b1\u03b9 \u03b4\u03b5\u03bd \u03b1\u03bd\u03c4\u03b9\u03c0\u03c1\u03bf\u03c3\u03c9\u03c0\u03b5\u03cd\u03bf\u03c5\u03bd \u03ba\u03ac\u03b8\u03b5 \u03c0\u03b9\u03b8\u03b1\u03bd\u03ae \u03bc\u03bf\u03c1\u03c6\u03ae \u03c4\u03c9\u03bd \u03b1\u03bd\u03c4\u03af\u03c3\u03c4\u03bf\u03b9\u03c7\u03c9\u03bd \u03b1\u03c1\u03c7\u03b9\u03c4\u03b5\u03ba\u03c4\u03bf\u03bd\u03b9\u03ba\u03ce\u03bd.<\/p>\n<p>\u0397 hosted Mem0 condition \u03b5\u03af\u03bd\u03b1\u03b9 \u03bc\u03af\u03b1 \u03b1\u03ba\u03c1\u03b9\u03b2\u03ae\u03c2 pinned literal-storage \u03b4\u03b9\u03b1\u03bc\u03cc\u03c1\u03c6\u03c9\u03c3\u03b7. \u0394\u03b5\u03bd \u03b4\u03b9\u03ba\u03b1\u03b9\u03bf\u03bb\u03bf\u03b3\u03b5\u03af claims \u03b3\u03b9\u03b1 \u03cc\u03bb\u03bf \u03c4\u03bf \u03c0\u03c1\u03bf\u03ca\u03cc\u03bd, \u03ac\u03bb\u03bb\u03b5\u03c2 \u03c1\u03c5\u03b8\u03bc\u03af\u03c3\u03b5\u03b9\u03c2 \u03ae \u03bc\u03b5\u03bb\u03bb\u03bf\u03bd\u03c4\u03b9\u03ba\u03ad\u03c2 \u03b5\u03ba\u03b4\u03cc\u03c3\u03b5\u03b9\u03c2. \u0397 \u03bc\u03b7 \u03b1\u03c0\u03cc\u03c1\u03c1\u03b9\u03c8\u03b7 \u03bc\u03b9\u03b1\u03c2 \u03b4\u03b9\u03b1\u03c6\u03bf\u03c1\u03ac\u03c2 \u03b4\u03b5\u03bd \u03c3\u03b7\u03bc\u03b1\u03af\u03bd\u03b5\u03b9 \u03b9\u03c3\u03bf\u03b4\u03c5\u03bd\u03b1\u03bc\u03af\u03b1, \u03b5\u03bd\u03ce \u03b1\u03c1\u03b9\u03b8\u03bc\u03b7\u03c4\u03b9\u03ba\u03ae \u03c5\u03c0\u03b5\u03c1\u03bf\u03c7\u03ae \u03c7\u03c9\u03c1\u03af\u03c2 confirmatory \u03c3\u03cd\u03b3\u03ba\u03c1\u03b9\u03c3\u03b7 \u03b4\u03b5\u03bd \u03b3\u03af\u03bd\u03b5\u03c4\u03b1\u03b9 \u03b3\u03b5\u03bd\u03b9\u03ba\u03ae \u03ba\u03b1\u03c4\u03ac\u03c4\u03b1\u03be\u03b7.<\/p>\n<p>\u039f\u03c1\u03b9\u03c3\u03bc\u03ad\u03bd\u03b5\u03c2 \u03b4\u03b9\u03b1\u03b3\u03bd\u03c9\u03c3\u03c4\u03b9\u03ba\u03ad\u03c2 \u03b5\u03c4\u03b9\u03ba\u03ad\u03c4\u03b5\u03c2 \u03b5\u03af\u03bd\u03b1\u03b9 \u03b5\u03c1\u03b3\u03b1\u03bb\u03b5\u03af\u03b1 \u03b1\u03bd\u03ac\u03bb\u03c5\u03c3\u03b7\u03c2 \u03ba\u03b1\u03b9 \u03cc\u03c7\u03b9 \u03ba\u03b1\u03b8\u03bf\u03bb\u03b9\u03ba\u03bf\u03af \u03bf\u03c1\u03b9\u03c3\u03bc\u03bf\u03af \u00ab\u03ba\u03b1\u03bb\u03ae\u03c2 \u03bc\u03bd\u03ae\u03bc\u03b7\u03c2\u00bb. \u039c\u03b9\u03b1 \u03b5\u03c0\u03b9\u03c7\u03b5\u03af\u03c1\u03b7\u03c3\u03b7 \u03c0\u03c1\u03ad\u03c0\u03b5\u03b9 \u03bd\u03b1 \u03c0\u03c1\u03bf\u03c3\u03b1\u03c1\u03bc\u03cc\u03c3\u03b5\u03b9 traps, oracles \u03ba\u03b1\u03b9 scopes \u03c3\u03c4\u03b9\u03c2 \u03b4\u03b9\u03ba\u03ad\u03c2 \u03c4\u03b7\u03c2 \u03b4\u03b9\u03b1\u03b4\u03b9\u03ba\u03b1\u03c3\u03af\u03b5\u03c2 \u03b1\u03bd\u03c4\u03af \u03bd\u03b1 \u03b1\u03bd\u03c4\u03b9\u03b3\u03c1\u03ac\u03c8\u03b5\u03b9 \u03bc\u03b7\u03c7\u03b1\u03bd\u03b9\u03ba\u03ac \u03c4\u03bf benchmark. \u0393\u03b9\u03b1 e-commerce \u03ae CRM, \u03b3\u03b9\u03b1 \u03c0\u03b1\u03c1\u03ac\u03b4\u03b5\u03b9\u03b3\u03bc\u03b1, \u03bc\u03b9\u03b1 \u03c0\u03c1\u03bf\u03c3\u03c9\u03c1\u03b9\u03bd\u03ae \u03b5\u03be\u03b1\u03af\u03c1\u03b5\u03c3\u03b7 \u03c0\u03b5\u03bb\u03ac\u03c4\u03b7 \u03ba\u03b1\u03b9 \u03bc\u03b9\u03b1 \u03bc\u03cc\u03bd\u03b9\u03bc\u03b7 \u03c0\u03bf\u03bb\u03b9\u03c4\u03b9\u03ba\u03ae \u03c7\u03c1\u03b5\u03b9\u03ac\u03b6\u03bf\u03bd\u03c4\u03b1\u03b9 \u03b4\u03b9\u03b1\u03c6\u03bf\u03c1\u03b5\u03c4\u03b9\u03ba\u03cc subject, scope, authority \u03ba\u03b1\u03b9 expiry.<\/p>\n<p>\u03a4\u03bf \u03c3\u03c4\u03c1\u03b1\u03c4\u03b7\u03b3\u03b9\u03ba\u03cc \u03c3\u03c5\u03bc\u03c0\u03ad\u03c1\u03b1\u03c3\u03bc\u03b1 \u03b5\u03af\u03bd\u03b1\u03b9 \u03ba\u03b1\u03b8\u03b1\u03c1\u03cc: \u03b7 \u03bc\u03bd\u03ae\u03bc\u03b7 \u03b5\u03bd\u03cc\u03c2 AI agent \u03b5\u03af\u03bd\u03b1\u03b9 \u03c3\u03cd\u03c3\u03c4\u03b7\u03bc\u03b1 \u03b1\u03c0\u03bf\u03c6\u03ac\u03c3\u03b5\u03c9\u03bd, \u03cc\u03c7\u03b9 \u03b1\u03c0\u03bb\u03ae \u03b1\u03c0\u03bf\u03b8\u03ae\u03ba\u03b7. \u0391\u03be\u03b9\u03cc\u03c0\u03b9\u03c3\u03c4\u03b7 \u03bb\u03b5\u03b9\u03c4\u03bf\u03c5\u03c1\u03b3\u03af\u03b1 \u03b1\u03c0\u03b1\u03b9\u03c4\u03b5\u03af \u03c0\u03c1\u03c9\u03c4\u03bf\u03b3\u03b5\u03bd\u03ae \u03b1\u03c0\u03bf\u03b4\u03b5\u03b9\u03ba\u03c4\u03b9\u03ba\u03ac \u03c3\u03c4\u03bf\u03b9\u03c7\u03b5\u03af\u03b1, typed \u03b1\u03bd\u03b1\u03c0\u03b1\u03c1\u03b1\u03c3\u03c4\u03ac\u03c3\u03b5\u03b9\u03c2 \u03bc\u03b5 provenance, \u03ba\u03b1\u03bd\u03cc\u03bd\u03b5\u03c2 supersession, \u03b5\u03bb\u03b5\u03b3\u03c7\u03cc\u03bc\u03b5\u03bd\u03b7 \u03b1\u03bd\u03ac\u03ba\u03c4\u03b7\u03c3\u03b7, \u03c0\u03b5\u03c1\u03b9\u03bf\u03c1\u03b9\u03c3\u03bc\u03ad\u03bd\u03b1 \u03b4\u03b9\u03ba\u03b1\u03b9\u03ce\u03bc\u03b1\u03c4\u03b1 \u03ba\u03b1\u03b9 tests \u03c0\u03bf\u03c5 \u03bc\u03b5\u03c4\u03c1\u03bf\u03cd\u03bd \u03c4\u03b7\u03bd \u03c4\u03b5\u03bb\u03b9\u03ba\u03ae \u03b5\u03bd\u03ad\u03c1\u03b3\u03b5\u03b9\u03b1. \u03a4\u03bf \u03ce\u03c1\u03b9\u03bc\u03bf \u03b5\u03c0\u03cc\u03bc\u03b5\u03bd\u03bf \u03b2\u03ae\u03bc\u03b1 \u03b4\u03b5\u03bd \u03b5\u03af\u03bd\u03b1\u03b9 \u03bd\u03b1 \u03b4\u03ce\u03c3\u03b5\u03c4\u03b5 \u03c3\u03c4\u03bf\u03bd agent \u00ab\u03cc\u03bb\u03b7 \u03c4\u03b7\u03bd \u03b9\u03c3\u03c4\u03bf\u03c1\u03af\u03b1\u00bb, \u03b1\u03bb\u03bb\u03ac \u03bd\u03b1 \u03bf\u03c1\u03af\u03c3\u03b5\u03c4\u03b5 \u03c0\u03bf\u03b9\u03b1 \u03b9\u03c3\u03c4\u03bf\u03c1\u03af\u03b1 \u03b8\u03b5\u03c9\u03c1\u03b5\u03af\u03c4\u03b1\u03b9 \u03ad\u03b3\u03ba\u03c5\u03c1\u03b7, \u03c0\u03cc\u03c4\u03b5 \u03c0\u03b1\u03cd\u03b5\u03b9 \u03bd\u03b1 \u03b9\u03c3\u03c7\u03cd\u03b5\u03b9 \u03ba\u03b1\u03b9 \u03c0\u03ce\u03c2 \u03b1\u03c0\u03bf\u03b4\u03b5\u03b9\u03ba\u03bd\u03cd\u03b5\u03c4\u03b1\u03b9 \u03cc\u03c4\u03b9 \u03c7\u03c1\u03b7\u03c3\u03b9\u03bc\u03bf\u03c0\u03bf\u03b9\u03ae\u03b8\u03b7\u03ba\u03b5 \u03c3\u03c9\u03c3\u03c4\u03ac.<\/p>\n<section class=\"td-service-cta\">\n<div class=\"td-service-cta-content\">\n<p class=\"td-service-cta-kicker\">\u0391\u03c0\u03cc \u03c4\u03b7 \u03bc\u03bd\u03ae\u03bc\u03b7 agent \u03c3\u03b5 \u03b5\u03bb\u03b5\u03b3\u03c7\u03cc\u03bc\u03b5\u03bd\u03bf \u03b5\u03c0\u03b9\u03c7\u03b5\u03b9\u03c1\u03b7\u03bc\u03b1\u03c4\u03b9\u03ba\u03cc workflow<\/p>\n<h3>\u03a3\u03c7\u03b5\u03b4\u03b9\u03ac\u03c3\u03c4\u03b5 provenance, retrieval gates, permissions \u03ba\u03b1\u03b9 \u03b1\u03bd\u03b5\u03be\u03ac\u03c1\u03c4\u03b7\u03c4\u03b7 \u03b5\u03c0\u03b1\u03bb\u03ae\u03b8\u03b5\u03c5\u03c3\u03b7<\/h3>\n<p>\u0397 TWO DOTS \u03c7\u03b1\u03c1\u03c4\u03bf\u03b3\u03c1\u03b1\u03c6\u03b5\u03af \u03c4\u03bf use case, \u03bf\u03c1\u03b3\u03b1\u03bd\u03ce\u03bd\u03b5\u03b9 \u03b1\u03c3\u03c6\u03b1\u03bb\u03ae pilots \u03ba\u03b1\u03b9 \u03c3\u03c5\u03bd\u03b4\u03ad\u03b5\u03b9 AI agents, \u03b5\u03c4\u03b1\u03b9\u03c1\u03b9\u03ba\u03ac \u03b4\u03b5\u03b4\u03bf\u03bc\u03ad\u03bd\u03b1, policy controls, monitoring \u03ba\u03b1\u03b9 \u03b1\u03bd\u03b8\u03c1\u03ce\u03c0\u03b9\u03bd\u03bf handoff \u03c0\u03c1\u03b9\u03bd \u03b7 \u03bc\u03bd\u03ae\u03bc\u03b7 \u03b1\u03c0\u03bf\u03ba\u03c4\u03ae\u03c3\u03b5\u03b9 \u03c0\u03c1\u03b1\u03b3\u03bc\u03b1\u03c4\u03b9\u03ba\u03ac side effects.<\/p>\n<div class=\"td-service-cta-actions\"><a class=\"td-service-cta-button\" href=\"https:\/\/twodots.gr\/aftomatismoi-epicheiriseon-ai\/\">Check out Business Automation &amp; AI<\/a><\/div>\n<\/div>\n<\/section>\n<section id=\"sychnes-erotiseis\" class=\"td-faq-section\">\n<div class=\"td-faq\">\n<p class=\"td-faq-heading\">Frequently Asked Questions (FAQs)<\/p>\n<details class=\"td-faq-item\">\n<summary class=\"td-faq-title\">\u03a4\u03b9 \u03b5\u03af\u03bd\u03b1\u03b9 \u03c4\u03bf DreamBench-SWE;<\/summary>\n<div class=\"td-faq-content\">\n<p>\u0395\u03af\u03bd\u03b1\u03b9 benchmark \u03c0\u03bf\u03bb\u03bb\u03b1\u03c0\u03bb\u03ce\u03bd \u03c3\u03c5\u03bd\u03b5\u03b4\u03c1\u03b9\u03ce\u03bd \u03b3\u03b9\u03b1 software agents. \u0395\u03bb\u03ad\u03b3\u03c7\u03b5\u03b9 \u03b1\u03bd \u03c3\u03c4\u03bf\u03b9\u03c7\u03b5\u03af\u03b1 \u03b1\u03c0\u03cc \u03c0\u03c1\u03bf\u03b7\u03b3\u03bf\u03cd\u03bc\u03b5\u03bd\u03b5\u03c2 \u03c3\u03c5\u03bd\u03b5\u03b4\u03c1\u03af\u03b5\u03c2 \u03c7\u03c1\u03b7\u03c3\u03b9\u03bc\u03bf\u03c0\u03bf\u03b9\u03bf\u03cd\u03bd\u03c4\u03b1\u03b9 \u03c3\u03c9\u03c3\u03c4\u03ac \u03c3\u03b5 \u03bc\u03b5\u03c4\u03b1\u03b3\u03b5\u03bd\u03ad\u03c3\u03c4\u03b5\u03c1\u03b5\u03c2 \u03b1\u03bb\u03bb\u03b1\u03b3\u03ad\u03c2 \u03ba\u03ce\u03b4\u03b9\u03ba\u03b1 \u03c0\u03bf\u03c5 \u03b2\u03b1\u03b8\u03bc\u03bf\u03bb\u03bf\u03b3\u03bf\u03cd\u03bd\u03c4\u03b1\u03b9 \u03bc\u03b5 \u03ba\u03c1\u03c5\u03c6\u03ac, \u03b5\u03ba\u03c4\u03b5\u03bb\u03ad\u03c3\u03b9\u03bc\u03b1 oracles.<\/p>\n<\/div>\n<\/details>\n<details class=\"td-faq-item\">\n<summary class=\"td-faq-title\">\u03a4\u03b9 \u03c3\u03b7\u03bc\u03b1\u03af\u03bd\u03b5\u03b9 memory hygiene;<\/summary>\n<div class=\"td-faq-content\">\n<p>\u03a3\u03b7\u03bc\u03b1\u03af\u03bd\u03b5\u03b9 \u03cc\u03c4\u03b9 \u03b7 \u03bc\u03bd\u03ae\u03bc\u03b7 \u03b4\u03b9\u03b1\u03c4\u03b7\u03c1\u03b5\u03af \u03c4\u03b5\u03ba\u03bc\u03b7\u03c1\u03b9\u03c9\u03bc\u03ad\u03bd\u03b7 \u03c0\u03bb\u03b7\u03c1\u03bf\u03c6\u03bf\u03c1\u03af\u03b1, \u03b1\u03c0\u03bf\u03c6\u03b5\u03cd\u03b3\u03b5\u03b9 stale \u03ae \u03ac\u03c3\u03c7\u03b5\u03c4\u03b1 \u03c3\u03c4\u03bf\u03b9\u03c7\u03b5\u03af\u03b1, \u03c3\u03ad\u03b2\u03b5\u03c4\u03b1\u03b9 \u03c4\u03bf scope, \u03b1\u03bd\u03b1\u03b3\u03bd\u03c9\u03c1\u03af\u03b6\u03b5\u03b9 supersession \u03ba\u03b1\u03b9 \u03b5\u03c0\u03b9\u03c4\u03c1\u03ad\u03c0\u03b5\u03b9 \u03c3\u03c4\u03bf\u03bd agent \u03bd\u03b1 \u03b1\u03c0\u03ad\u03c7\u03b5\u03b9 \u03cc\u03c4\u03b1\u03bd \u03bf\u03b9 \u03b1\u03c0\u03bf\u03b4\u03b5\u03af\u03be\u03b5\u03b9\u03c2 \u03b4\u03b5\u03bd \u03b5\u03c0\u03b1\u03c1\u03ba\u03bf\u03cd\u03bd.<\/p>\n<\/div>\n<\/details>\n<details class=\"td-faq-item\">\n<summary class=\"td-faq-title\">\u0391\u03c0\u03ad\u03b4\u03b5\u03b9\u03be\u03b5 \u03b7 \u03bc\u03b5\u03bb\u03ad\u03c4\u03b7 \u03cc\u03c4\u03b9 \u03b7 Mem0 \u03b5\u03af\u03bd\u03b1\u03b9 \u03b7 \u03ba\u03b1\u03bb\u03cd\u03c4\u03b5\u03c1\u03b7 \u03bb\u03cd\u03c3\u03b7;<\/summary>\n<div class=\"td-faq-content\">\n<p>\u038c\u03c7\u03b9. \u039c\u03af\u03b1 \u03c3\u03c5\u03b3\u03ba\u03b5\u03ba\u03c1\u03b9\u03bc\u03ad\u03bd\u03b7 pinned hosted literal-storage \u03c1\u03cd\u03b8\u03bc\u03b9\u03c3\u03b7 \u03c0\u03ad\u03c4\u03c5\u03c7\u03b5 97\/180 \u03c3\u03c4\u03bf v2.1, \u03b1\u03bb\u03bb\u03ac \u03b7 \u03bc\u03b5\u03bb\u03ad\u03c4\u03b7 \u03c0\u03b5\u03c1\u03b9\u03bf\u03c1\u03af\u03b6\u03b5\u03b9 \u03c4\u03bf claim \u03c3\u03b5 \u03b1\u03c5\u03c4\u03cc \u03c4\u03bf configuration \u03ba\u03b1\u03b9 \u03b4\u03b5\u03bd \u03c4\u03b5\u03ba\u03bc\u03b7\u03c1\u03b9\u03ce\u03bd\u03b5\u03b9 \u03b3\u03b5\u03bd\u03b9\u03ba\u03ae \u03c5\u03c0\u03b5\u03c1\u03bf\u03c7\u03ae \u03c0\u03c1\u03bf\u03ca\u03cc\u03bd\u03c4\u03bf\u03c2 \u03ae \u03bc\u03b7\u03c7\u03b1\u03bd\u03b9\u03c3\u03bc\u03bf\u03cd.<\/p>\n<\/div>\n<\/details>\n<details class=\"td-faq-item\">\n<summary class=\"td-faq-title\">\u0389\u03c4\u03b1\u03bd \u03c4\u03bf typed-plus-raw \u03c3\u03cd\u03c3\u03c4\u03b7\u03bc\u03b1 \u03ba\u03b1\u03bb\u03cd\u03c4\u03b5\u03c1\u03bf \u03b1\u03c0\u03cc \u03c4\u03bf verbatim memory;<\/summary>\n<div class=\"td-faq-content\">\n<p>\u03a3\u03c4\u03bf v2 \u03b5\u03af\u03c7\u03b5 95\/180 \u03ad\u03bd\u03b1\u03bd\u03c4\u03b9 89\/180, \u03b1\u03bb\u03bb\u03ac \u03b7 \u03c3\u03c4\u03b1\u03c4\u03b9\u03c3\u03c4\u03b9\u03ba\u03ae \u03c3\u03cd\u03b3\u03ba\u03c1\u03b9\u03c3\u03b7 \u03b4\u03b5\u03bd \u03b1\u03c0\u03ad\u03c1\u03c1\u03b9\u03c8\u03b5 \u03c4\u03b7 \u03bc\u03b7\u03b4\u03b5\u03bd\u03b9\u03ba\u03ae \u03c5\u03c0\u03cc\u03b8\u03b5\u03c3\u03b7. \u03a3\u03c4\u03bf v2.1 \u03bf\u03b9 \u03b5\u03c0\u03b9\u03c4\u03c5\u03c7\u03af\u03b5\u03c2 \u03ae\u03c4\u03b1\u03bd 83\/180 \u03ba\u03b1\u03b9 82\/180 \u03b1\u03bd\u03c4\u03af\u03c3\u03c4\u03bf\u03b9\u03c7\u03b1. \u0391\u03c5\u03c4\u03ac \u03b4\u03b5\u03bd \u03b1\u03c0\u03bf\u03b4\u03b5\u03b9\u03ba\u03bd\u03cd\u03bf\u03c5\u03bd \u03c5\u03c0\u03b5\u03c1\u03bf\u03c7\u03ae \u03ae \u03b9\u03c3\u03bf\u03b4\u03c5\u03bd\u03b1\u03bc\u03af\u03b1.<\/p>\n<\/div>\n<\/details>\n<details class=\"td-faq-item\">\n<summary class=\"td-faq-title\">\u0393\u03b9\u03b1\u03c4\u03af \u03c7\u03c1\u03b7\u03c3\u03b9\u03bc\u03bf\u03c0\u03bf\u03b9\u03bf\u03cd\u03bd\u03c4\u03b1\u03b9 hidden executable oracles;<\/summary>\n<div class=\"td-faq-content\">\n<p>\u0393\u03b9\u03b1 \u03bd\u03b1 \u03bc\u03b5\u03c4\u03c1\u03b7\u03b8\u03b5\u03af \u03b1\u03bd \u03b7 \u03bc\u03bd\u03ae\u03bc\u03b7 \u03bf\u03b4\u03ae\u03b3\u03b7\u03c3\u03b5 \u03c3\u03b5 \u03bb\u03b5\u03b9\u03c4\u03bf\u03c5\u03c1\u03b3\u03b9\u03ba\u03ac \u03c3\u03c9\u03c3\u03c4\u03ae \u03b1\u03bb\u03bb\u03b1\u03b3\u03ae \u03ba\u03ce\u03b4\u03b9\u03ba\u03b1 \u03ba\u03b1\u03b9 \u03cc\u03c7\u03b9 \u03b1\u03c0\u03bb\u03ce\u03c2 \u03c3\u03b5 \u03c0\u03b5\u03b9\u03c3\u03c4\u03b9\u03ba\u03ae \u03b3\u03bb\u03c9\u03c3\u03c3\u03b9\u03ba\u03ae \u03b1\u03c0\u03ac\u03bd\u03c4\u03b7\u03c3\u03b7. \u03a4\u03bf \u03ba\u03c1\u03c5\u03c6\u03cc oracle \u03bc\u03b5\u03b9\u03ce\u03bd\u03b5\u03b9 \u03c4\u03bf\u03bd \u03ba\u03af\u03bd\u03b4\u03c5\u03bd\u03bf \u03ac\u03bc\u03b5\u03c3\u03b7\u03c2 \u03c0\u03c1\u03bf\u03c3\u03b1\u03c1\u03bc\u03bf\u03b3\u03ae\u03c2 \u03c3\u03c4\u03b7 \u03bb\u03cd\u03c3\u03b7.<\/p>\n<\/div>\n<\/details>\n<details class=\"td-faq-item\">\n<summary class=\"td-faq-title\">\u03a0\u03bf\u03b9\u03b1 \u03b5\u03af\u03bd\u03b1\u03b9 \u03b7 \u03b1\u03be\u03af\u03b1 \u03c4\u03c9\u03bd raw episodes;<\/summary>\n<div class=\"td-faq-content\">\n<p>\u0394\u03b9\u03b1\u03c4\u03b7\u03c1\u03bf\u03cd\u03bd \u03c4\u03b7\u03bd \u03c0\u03c1\u03c9\u03c4\u03bf\u03b3\u03b5\u03bd\u03ae \u03b1\u03c0\u03cc\u03b4\u03b5\u03b9\u03be\u03b7 \u03c0\u03af\u03c3\u03c9 \u03b1\u03c0\u03cc \u03ba\u03ac\u03b8\u03b5 derived \u03bc\u03bd\u03ae\u03bc\u03b7. \u0388\u03c4\u03c3\u03b9 \u03b7 \u03bf\u03bc\u03ac\u03b4\u03b1 \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b5\u03bb\u03ad\u03b3\u03be\u03b5\u03b9 \u03bc\u03b9\u03b1 \u03c3\u03cd\u03bd\u03bf\u03c8\u03b7, \u03bd\u03b1 \u03b4\u03b9\u03bf\u03c1\u03b8\u03ce\u03c3\u03b5\u03b9 \u03bb\u03ac\u03b8\u03bf\u03c2 consolidation \u03ba\u03b1\u03b9 \u03bd\u03b1 \u03ba\u03c1\u03b1\u03c4\u03ae\u03c3\u03b5\u03b9 audit trail \u03c7\u03c9\u03c1\u03af\u03c2 \u03bd\u03b1 \u03c7\u03ac\u03c3\u03b5\u03b9 \u03c4\u03bf \u03b1\u03c1\u03c7\u03b9\u03ba\u03cc \u03b3\u03b5\u03b3\u03bf\u03bd\u03cc\u03c2.<\/p>\n<\/div>\n<\/details>\n<details class=\"td-faq-item\">\n<summary class=\"td-faq-title\">\u039c\u03c0\u03bf\u03c1\u03b5\u03af \u03c4\u03bf benchmark \u03bd\u03b1 \u03b5\u03c6\u03b1\u03c1\u03bc\u03bf\u03c3\u03c4\u03b5\u03af \u03b5\u03ba\u03c4\u03cc\u03c2 software engineering;<\/summary>\n<div class=\"td-faq-content\">\n<p>\u0397 \u03bc\u03b5\u03bb\u03ad\u03c4\u03b7 \u03b4\u03b5\u03bd \u03b1\u03c0\u03bf\u03b4\u03b5\u03b9\u03ba\u03bd\u03cd\u03b5\u03b9 \u03ac\u03bc\u03b5\u03c3\u03b7 \u03bc\u03b5\u03c4\u03b1\u03c6\u03bf\u03c1\u03ac. \u039f\u03b9 \u03b1\u03c1\u03c7\u03ad\u03c2 provenance, scope, supersession \u03ba\u03b1\u03b9 executable validation \u03bc\u03c0\u03bf\u03c1\u03bf\u03cd\u03bd \u03bd\u03b1 \u03ba\u03b1\u03b8\u03bf\u03b4\u03b7\u03b3\u03ae\u03c3\u03bf\u03c5\u03bd \u03b5\u03c4\u03b1\u03b9\u03c1\u03b9\u03ba\u03ac tests, \u03b1\u03bb\u03bb\u03ac \u03c7\u03c1\u03b5\u03b9\u03ac\u03b6\u03bf\u03bd\u03c4\u03b1\u03b9 traps \u03ba\u03b1\u03b9 oracles \u03b5\u03b9\u03b4\u03b9\u03ba\u03ac \u03b3\u03b9\u03b1 \u03c4\u03bf \u03b5\u03ba\u03ac\u03c3\u03c4\u03bf\u03c4\u03b5 workflow.<\/p>\n<\/div>\n<\/details>\n<details class=\"td-faq-item\">\n<summary class=\"td-faq-title\">\u03a4\u03b9 \u03c0\u03c1\u03ad\u03c0\u03b5\u03b9 \u03bd\u03b1 \u03bc\u03b5\u03c4\u03c1\u03ae\u03c3\u03b5\u03b9 \u03c0\u03c1\u03ce\u03c4\u03b1 \u03bc\u03b9\u03b1 \u03b5\u03c0\u03b9\u03c7\u03b5\u03af\u03c1\u03b7\u03c3\u03b7;<\/summary>\n<div class=\"td-faq-content\">\n<p>\u039c\u03b1\u03b6\u03af \u03bc\u03b5 \u03c4\u03bf task success \u03c0\u03c1\u03ad\u03c0\u03b5\u03b9 \u03bd\u03b1 \u03ba\u03b1\u03c4\u03b1\u03b3\u03c1\u03ac\u03c6\u03b5\u03b9 stale use, overscope, irrelevant retrieval, \u03b5\u03c0\u03b1\u03bd\u03ac\u03bb\u03b7\u03c8\u03b7 \u03c0\u03b1\u03bb\u03b9\u03ce\u03bd \u03bb\u03b1\u03b8\u03ce\u03bd, \u03b4\u03c5\u03bd\u03b1\u03c4\u03cc\u03c4\u03b7\u03c4\u03b1 \u03b1\u03c0\u03bf\u03c7\u03ae\u03c2, latency, \u03ba\u03cc\u03c3\u03c4\u03bf\u03c2 \u03b1\u03bd\u03ac \u03b5\u03c0\u03b9\u03c4\u03c5\u03c7\u03ae \u03b5\u03c1\u03b3\u03b1\u03c3\u03af\u03b1 \u03ba\u03b1\u03b9 auditability.<\/p>\n<\/div>\n<\/details>\n<\/div>\n<\/section>\n<div class=\"td-source-list\">\n<p id=\"piges\" class=\"td-source-list-title\">Sources<\/p>\n<ul>\n<li><a href=\"https:\/\/arxiv.org\/abs\/2608.20664\" target=\"_blank\" rel=\"noopener\">arXiv \u2014 DreamBench-SWE: A Multi-Session Memory-Hygiene Benchmark for Software Agents<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/iroiro147\/dreambench-swe\/releases\/tag\/v2.1.0\" target=\"_blank\" rel=\"noopener\">GitHub \u2014 DreamBench-SWE v2.1.0 benchmark and evidence release<\/a><\/li>\n<li><a href=\"https:\/\/www.nist.gov\/publications\/artificial-intelligence-risk-management-framework-generative-artificial-intelligence\" target=\"_blank\" rel=\"noopener\">NIST \u2014 AI RMF Generative AI Profile<\/a><\/li>\n<li><a href=\"https:\/\/genai.owasp.org\/2026\/05\/13\/memory-is-a-feature-it-is-also-an-attack-surface\/\" target=\"_blank\" rel=\"noopener\">OWASP \u2014 Memory Is a Feature. It Is Also an Attack Surface<\/a><\/li>\n<\/ul>\n<\/div>","protected":false},"excerpt":{"rendered":"<p>\u03a4\u03bf DreamBench-SWE \u03b4\u03b5\u03af\u03c7\u03bd\u03b5\u03b9 \u03b3\u03b9\u03b1\u03c4\u03af \u03b7 \u03bc\u03bd\u03ae\u03bc\u03b7 AI agents \u03c7\u03c1\u03b5\u03b9\u03ac\u03b6\u03b5\u03c4\u03b1\u03b9 provenance, \u03c3\u03c9\u03c3\u03c4\u03cc scope, retrieval gates \u03ba\u03b1\u03b9 \u03b5\u03ba\u03c4\u03b5\u03bb\u03ad\u03c3\u03b9\u03bc\u03b7 \u03b5\u03c0\u03b1\u03bb\u03ae\u03b8\u03b5\u03c5\u03c3\u03b7.<\/p>","protected":false},"author":1,"featured_media":90743,"comment_status":"open","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"_gspb_post_css":"","content-type":"","footnotes":""},"categories":[199],"tags":[7231,6506,17818,19846,19845],"class_list":["post-90741","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-techniti-noimosyni","tag-agentic-ai","tag-ai-agents","tag-ai-benchmarks","tag-software-engineering","tag-mnimi-ai"],"blocksy_meta":{"styles_descriptor":{"styles":{"desktop":"","tablet":"","mobile":""},"google_fonts":[],"version":7}},"_links":{"self":[{"href":"https:\/\/twodots.gr\/en\/wp-json\/wp\/v2\/posts\/90741","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/twodots.gr\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/twodots.gr\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/twodots.gr\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/twodots.gr\/en\/wp-json\/wp\/v2\/comments?post=90741"}],"version-history":[{"count":1,"href":"https:\/\/twodots.gr\/en\/wp-json\/wp\/v2\/posts\/90741\/revisions"}],"predecessor-version":[{"id":90744,"href":"https:\/\/twodots.gr\/en\/wp-json\/wp\/v2\/posts\/90741\/revisions\/90744"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/twodots.gr\/en\/wp-json\/wp\/v2\/media\/90743"}],"wp:attachment":[{"href":"https:\/\/twodots.gr\/en\/wp-json\/wp\/v2\/media?parent=90741"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/twodots.gr\/en\/wp-json\/wp\/v2\/categories?post=90741"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/twodots.gr\/en\/wp-json\/wp\/v2\/tags?post=90741"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}