
{"id":10734,"date":"2026-08-04T18:53:04","date_gmt":"2026-08-04T10:53:04","guid":{"rendered":"https:\/\/infernews.com\/blog\/the-long-horizon-computer-use-harness-run-ai-agents-across-desktop-apps-and-the\/"},"modified":"2026-08-04T18:59:15","modified_gmt":"2026-08-04T10:59:15","slug":"the-long-horizon-computer-use-harness-run-ai-agents-across-desktop-apps-and-the","status":"publish","type":"post","link":"https:\/\/infernews.com\/blog\/the-long-horizon-computer-use-harness-run-ai-agents-across-desktop-apps-and-the\/","title":{"rendered":"LongHorizon-Harness \u89e3\u6c7a\u4ee3\u7406\u5728\u9577\u4efb\u52d9\u6642\u7684\u932f\u8aa4"},"content":{"rendered":"\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/infernews.com\/blog\/wp-content\/uploads\/2026\/08\/quickstart.gif\" alt=\"Install and run LongHorizon-Harness from the command line\"\/><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">\u7576\u4ee3\u7406\u8981\u6a6b\u8de8\u684c\u9762\u7a0b\u5f0f\u540c command line \u9023\u7e8c\u505a\u4e8b\uff0c\u6700\u6613\u51fa\u554f\u984c\u5514\u4fc2\u55ae\u6b65\u64cd\u4f5c\uff0c\u800c\u4fc2\u4e2d\u9014\u8a18\u932f\u72c0\u614b\u3001\u5224\u65b7\u932f\u9032\u5ea6\uff0c\u7d50\u679c\u6108\u505a\u6108\u504f\u3002LongHorizon-Harness \u91dd\u5c0d\u5c6c\u65bc<strong>\u9577\u6642\u7a0b\u4ee3\u7406\u57f7\u884c\u8207\u9a57\u8b49\u5de5\u5177<\/strong>\uff0c\u76ee\u6a19\u4fc2\u4ee4\u8907\u96dc\u4efb\u52d9\u53ef\u4ee5\u88ab\u4fdd\u5b58\u3001\u6838\u5be6\uff0c\u518d\u4e00\u8def\u63a8\u9032\u5230\u5b8c\u6210\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u5b83\u5514\u4fc2\u65b0\u6a21\u578b\uff0c\u5b83\u4e3b\u8981\u5354\u52a9 Claude Code\u3001Codex\u3001OpenClaw \u4e4b\u985e agent backend \u4e0a\u9762\u8655\u7406 execution\u3001state management \u540c result verification\u3002\u6838\u5fc3\u505a\u6cd5\u662f Manage-Execute-Audit (MEA) loop\uff0c\u628a\u898f\u5283\u3001\u57f7\u884c\u3001\u5be9\u6838\u62c6\u6210\u4e0d\u540c\u8def\u5f91\uff0c\u4e26\u628a\u53ef\u4fe1\u72c0\u614b\u7368\u7acb\u4fdd\u5b58\uff0c\u6e1b\u5c11\u55ae\u4e00\u9577\u5c0d\u8a71\u6108\u6efe\u6108\u4e82\u7684\u554f\u984c\u3002<\/p>\n\n\n\n<figure class=\"wp-block-image size-full\"><img data-dominant-color=\"cbcacb\" data-has-transparency=\"false\" style=\"--dominant-color: #cbcacb;\" loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"682\" src=\"https:\/\/infernews.com\/blog\/wp-content\/uploads\/2026\/08\/image.png\" alt=\"\" class=\"wp-image-10738 not-transparent\" srcset=\"https:\/\/infernews.com\/blog\/wp-content\/uploads\/2026\/08\/image.png 1024w, https:\/\/infernews.com\/blog\/wp-content\/uploads\/2026\/08\/image-300x200.png 300w, https:\/\/infernews.com\/blog\/wp-content\/uploads\/2026\/08\/image-768x512.png 768w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u652f\u63f4 Claude Code\u3001Codex\u3001OpenClaw\uff0c\u4ea6\u53ef\u914d Gemini CLI \u8207 mini-SWE-agent<\/li>\n\n\n\n<li>\u4ee5\u7368\u7acb audited state \u63a8\u52d5\u4e0b\u4e00\u6b65\uff0c\u800c\u5514\u4fc2\u53ea\u9760 session \u5167\u8a18\u61b6<\/li>\n\n\n\n<li>\u53ef\u7528\u55ae\u4e00\u6307\u4ee4\u555f\u52d5\uff0c\u6bcf\u6b21\u57f7\u884c\u6703\u7368\u7acb\u4fdd\u7559 audit trail<\/li>\n\n\n\n<li>\u91dd\u5c0d WeaveBench\u3001OSWorld 2.0\u3001Terminal-Bench 2.1 \u9019\u985e\u9577\u4efb\u52d9\u57fa\u6e96\u6709\u516c\u958b\u6210\u7e3e<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">\u5b83\u5728 WeaveBench \u53d6\u5f97 80.7% PassRate\u3001OSWorld 2.0 \u6709 35.2% partial score\uff0cTerminal-Bench 2.1 success rate \u70ba 77.2%\u3002\u5b98\u65b9\u4ea6\u5f37\u8abf\u5728\u76f8\u540c backbone \u4e0b\uff0c\u53ea\u63db harness\uff0c\u4e09\u500b benchmark \u90fd\u5411\u4e0a\u8d70\uff0c\u53cd\u6620\u6539\u9032\u9ede\u4e3b\u8981\u4f86\u81ea\u6d41\u7a0b\u63a7\u5236\u540c\u9a57\u8b49\u6a5f\u5236\uff0c\u800c\u5514\u4fc2\u6a21\u578b\u7a81\u7136\u8b8a\u5f37\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u5462\u7a2e\u8a2d\u8a08\u8f03\u9069\u5408\u9700\u8981\u9577\u6642\u9593\u81ea\u52d5\u5316\u8655\u7406\u3001\u591a\u6b65\u9a5f\u4ea4\u63a5\u3001\u53c8\u8981\u8ffd\u8e64\u7d50\u679c\u53ef\u4fe1\u5ea6\u7684\u5718\u968a\uff0c\u4f8b\u5982\u7814\u7a76\u3001\u8edf\u4ef6\u6e2c\u8a66\u3001\u7cfb\u7d71\u64cd\u4f5c\u540c\u8907\u96dc office workflow\u3002\u4ee3\u50f9\u662f\u6d41\u7a0b\u6bd4\u55ae\u7d14\u804a\u5929\u5f0f\u4ee3\u7406\u66f4\u91cd\uff0c\u5be9\u6838\u8207\u72c0\u614b\u7ba1\u7406\u6703\u589e\u52a0\u7d50\u69cb\u8207\u6210\u672c\uff0c\u4f46\u63db\u4f86\u7684\u662f\u66f4\u7a69\u5b9a\u7684\u5ef6\u7e8c\u80fd\u529b\uff0c\u540c\u8f03\u5bb9\u6613\u8ffd\u67e5\u6bcf\u4e00\u6b65\u9ede\u6a23\u505a\u51fa\u4f86\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/lh-harness.pages.dev\/\" rel=\"noopener noreferrer\"><strong>\u9805\u76ee\u4e3b\u9801<\/strong><\/a> \u00b7 <a href=\"https:\/\/github.com\/AMAP-ML\/LongHorizon-Harness\" rel=\"noopener noreferrer\"><strong>GitHub<\/strong><\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u4ee3\u7406\u80fd\u5426\u8dd1\u8db3\u5e7e\u5341\u5c0f\u6642\uff0c\u95dc\u9375\u5514\u53ea\u5728\u6a21\u578b\u80fd\u529b\uff0c\u4ef2\u5728\u72c0\u614b\u6709\u5187\u8d70\u6a23\u3002LongHorizon-Harness\u628a\u9a57\u8b49\u3001\u57f7\u884c\u540c\u4efb\u52d9\u5ef6\u7e8c\u62c6\u958b\u8655\u7406\u3002<\/p>\n","protected":false},"author":8,"featured_media":10733,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"ai_generated_summary":"","footnotes":""},"categories":[133,185,140,147,153,116,191,190,197,199,201,203],"tags":[],"class_list":["post-10734","post","type-post","status-publish","format-standard","hentry","category-133","category-qwen","category-gemini","category-deepseek","category-openai","category-agentic","category-anthropic","category-openclaw","category-framework","category-dataset-","category-skill-","category-minimax"],"_links":{"self":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/10734","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/users\/8"}],"replies":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/comments?post=10734"}],"version-history":[{"count":6,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/10734\/revisions"}],"predecessor-version":[{"id":10744,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/10734\/revisions\/10744"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media\/10733"}],"wp:attachment":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media?parent=10734"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/categories?post=10734"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/tags?post=10734"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}