{"id":6118,"date":"2026-10-19T16:49:02","date_gmt":"2026-10-19T16:49:02","guid":{"rendered":"https:\/\/i10xnews.kinsta.cloud\/?p=6118"},"modified":"2026-10-19T17:17:21","modified_gmt":"2026-10-19T17:17:21","slug":"harnessdev-llms-build-own-agent-harness","status":"publish","type":"post","link":"https:\/\/i10x.ai\/news\/harnessdev-llms-build-own-agent-harness","title":{"rendered":"LLMs Can Build Their Own Agent Harnesses \u2014 But Still Trail Human Ones on Code"},"content":{"rendered":"\n<h2 class=\"wp-block-heading\">\u26a1 Quick Take<\/h2>\n\n<ul class=\"wp-block-list\"><li><strong>The finding:<\/strong> With identical weights, GPT-5 solves 35.2% of Terminal-Bench 2.1 in Terminus 2 but 49.6% in Codex CLI \u2014 the harness is half the story.<\/li><li><strong>Creation results:<\/strong> LLM-built harnesses still lag mature human references on code and search\/research, while matching or exceeding them on writing and ML experimentation.<\/li><li><strong>Evolution helps modestly:<\/strong> Self-runtime revisions improve held-out scores by +1.43 to +4.44 points (mean +3.11), but feedback scores predict held-out gains only 53.1% of the time.<\/li><\/ul>\n\n\n<p class=\"wp-block-paragraph\">Agent benchmarks usually fix a harness and score the model. ByteDance Seed and collaborators flip the question: can the model create and evolve the harness itself? HarnessDev grades runnable infrastructure, not just task answers.<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1344\" height=\"768\" src=\"https:\/\/i10xnews.kinsta.cloud\/wp-content\/uploads\/2026\/10\/harnessdev-llms-build-own-agent-harness-harness-effect.webp\" alt=\"Dumbbell: GPT-5 solves 35.2 percent of Terminal-Bench 2.1 in Terminus 2 versus 49.6 percent in Codex CLI.\" class=\"wp-image-6117\" srcset=\"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/harnessdev-llms-build-own-agent-harness-harness-effect.webp 1344w, https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/harnessdev-llms-build-own-agent-harness-harness-effect-300x171.webp 300w, https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/harnessdev-llms-build-own-agent-harness-harness-effect-1024x585.webp 1024w, https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/harnessdev-llms-build-own-agent-harness-harness-effect-768x439.webp 768w\" sizes=\"auto, (max-width: 1344px) 100vw, 1344px\" \/><figcaption class=\"wp-element-caption\">i10X chart based on data from Wu et al. (2026), arXiv:2609.01437, introductory harness comparison (CC BY-NC-ND 4.0).<\/figcaption><\/figure>\n\n\n\n<h2 class=\"wp-block-heading\">\ud83d\udd2c What the study shows<\/h2>\n\n\n<p class=\"wp-block-paragraph\">Creation starts from a minimal seed and a few cases; Evolution revises the agent&#8217;s own harness using downstream feedback. Results cover six creator LLMs, four domains and five downstream benchmarks totaling 2,207 unique instances, with hidden evaluation tasks withheld from development.<\/p>\n\n\n<ul class=\"wp-block-list\"><li><strong>Harness defects dominate some failures:<\/strong> 77.8% of failed Data tasks were attributed to harness defects rather than executor capability alone.<\/li><li><strong>Executor\u2013harness coupling:<\/strong> Switching executors can break specialized rules: in one Opus Search harness, duplicate-query rate jumped from 10.1% to 88.2% after the executor changed.<\/li><li><strong>Weak executors bottleneck strong harnesses:<\/strong> Several Qwen and DeepSeek harnesses improved under Gemini, including +17.6 on BrowseComp and +12.9 on MLE-bench.<\/li><li><strong>Diagnosis is rare:<\/strong> Trajectory inspection covered only 0.5%\u201340.2% of feedback tasks depending on lineage.<\/li><\/ul>\n\n\n<h2 class=\"wp-block-heading\">\u2705 Why it matters: the missing eval axis<\/h2>\n\n\n<p class=\"wp-block-paragraph\">If changing the harness moves scores by double digits, reporting only model\u2013task pairs is incomplete. HarnessDev makes infrastructure a first-class deliverable.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">\ud83e\uddea Method in brief<\/h2>\n\n\n<p class=\"wp-block-paragraph\">Capability (held-out task success) and efficiency (execution-token cost) scored for each constructed harness. Evolution uses feedback pairs from SWE-Pro-100 and Terminal-Bench-89.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">\u2753 What remains open<\/h2>\n\n<ul class=\"wp-block-list\"><li><strong>Reference ceilings:<\/strong> Exceeding 100% means beating the selected human reference, not humans in general.<\/li><li><strong>Feedback \u2260 held-out:<\/strong> Only 2 of 9 declared versions were held-out optimal.<\/li><\/ul>\n\n\n<h2 class=\"wp-block-heading\">\ud83e\udded Context<\/h2>\n\n\n<p class=\"wp-block-paragraph\">Related harness work includes HarnessAudit. See <a href=\"\/news\/category\/studies\/agent-research\">Agents<\/a>.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">\ud83d\udee0\ufe0f What it means for agent platforms<\/h2>\n\n<ul class=\"wp-block-list\"><li><strong>Version harnesses like models:<\/strong> They move the score.<\/li><li><strong>Re-test after executor swaps:<\/strong> Rules are model-specific.<\/li><li><strong>Inspect failures:<\/strong> Diagnosis is underused.<\/li><\/ul>\n\n\n<h2 class=\"wp-block-heading\">\u270d\ufe0f About the analysis<\/h2>\n\n\n<p class=\"wp-block-paragraph\">This article is based on the full text of the preprint &#8220;HarnessDev: Can LLMs Create and Evolve Their Own Agent Harness?&#8221; (arXiv:2609.01437v1, submitted September 1, 2026). All numbers are taken from the paper&#8217;s text, tables and figures; i10X did not rerun the experiments. The chart was redrawn by i10X from introductory harness comparison of the paper, which is published under a CC BY-NC-ND 4.0 license.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">\ud83d\udd2d i10x Perspective<\/h2>\n\n\n<p class=\"wp-block-paragraph\">HarnessDev\u2019s quiet provocation is that agents should be graded on the systems they build to run themselves. The 14-point Terminus-to-Codex gap for the same GPT-5 weights is the punchline: we have been scoring the wrong object.<\/p>\n\n","protected":false},"excerpt":{"rendered":"<p>HarnessDev: six creator models build execution infrastructure from a seed. Generated harnesses match or beat references on writing and ML experiments, but lag on code and search. Evolution adds a mean +3.11 points.<\/p>\n","protected":false},"author":2,"featured_media":6268,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"paper_title":"HarnessDev: Can LLMs Create and Evolve Their Own Agent Harness?","paper_url":"https:\/\/arxiv.org\/abs\/2609.01437","pdf_url":"https:\/\/arxiv.org\/pdf\/2609.01437v1","doi":"","arxiv_id":"2609.01437","authors":"Yuhao Wu, Jingyuan Zhang, Jiajun Shi, Xinping Lei, Qingshui Gu, Yuxuan Zhang, Zexuan Wang, Chen He, Chen Huang, Maojia Song, Zhiyuan Zeng, Shaowen Wang et al. (19 authors)","institutions":"ByteDance Seed; SUTD; Georgia Tech; M-A-P; TokenWave.AI","paper_date":"2026-09-01","venue":"arXiv (cs.SE), version 1","peer_reviewed":false,"code_url":"","data_url":"","funding_coi":"ByteDance Seed affiliation.","figure_license":"Chart drawn by i10X from introductory harness comparison (CC BY-NC-ND 4.0)","paper_license":"CC BY-NC-ND 4.0","replication":"no","study_models":"Six creator LLMs; downstream executors incl. GPT-5, Gemini, Qwen, DeepSeek, Opus","study_sample":"2,207 unique downstream instances across 5 benchmarks; Creation + Evolution stages","footnotes":"","_yoast_wpseo_primary_category":""},"categories":[55,52],"tags":[],"companies":[39,37,38],"products":[],"class_list":["post-6118","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-agent-research","category-studies","company-anthropic","company-google","company-open-ai"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v28.7 - https:\/\/yoast.com\/product\/yoast-seo-wordpress\/ -->\n<title>HarnessDev: Can LLMs Create Their Own Agent Harness? - i10X News<\/title>\n<meta name=\"description\" content=\"HarnessDev benchmark: LLM-built harnesses trail human ones on code\/search, match on writing\/ML. Evolution gains +3.11 points held-out.\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/i10xnews.kinsta.cloud\/harnessdev-llms-build-own-agent-harness\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"HarnessDev: Can LLMs Create Their Own Agent Harness? - i10X News\" \/>\n<meta property=\"og:description\" content=\"HarnessDev benchmark: LLM-built harnesses trail human ones on code\/search, match on writing\/ML. Evolution gains +3.11 points held-out.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/i10xnews.kinsta.cloud\/harnessdev-llms-build-own-agent-harness\" \/>\n<meta property=\"og:site_name\" content=\"i10X News\" \/>\n<meta property=\"article:published_time\" content=\"2026-10-19T16:49:02+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2026-10-19T17:17:21+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/harnessdev-llms-build-own-agent-harness-title.webp\" \/>\n\t<meta property=\"og:image:width\" content=\"1344\" \/>\n\t<meta property=\"og:image:height\" content=\"768\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/webp\" \/>\n<meta name=\"author\" content=\"Christopher Ort\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:creator\" content=\"@i10x_ai\" \/>\n<meta name=\"twitter:site\" content=\"@i10x_ai\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"Christopher Ort\" \/>\n\t<meta name=\"twitter:label2\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"2 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"NewsArticle\",\"@id\":\"https:\\\/\\\/i10xnews.kinsta.cloud\\\/harnessdev-llms-build-own-agent-harness#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/i10xnews.kinsta.cloud\\\/harnessdev-llms-build-own-agent-harness\"},\"author\":{\"name\":\"Christopher Ort\",\"@id\":\"https:\\\/\\\/news.i10x.ai\\\/#\\\/schema\\\/person\\\/d410bb7ee72febefca2c9c369939219a\"},\"headline\":\"LLMs Can Build Their Own Agent Harnesses \u2014 But Still Trail Human Ones on Code\",\"datePublished\":\"2026-10-19T16:49:02+00:00\",\"dateModified\":\"2026-10-19T17:17:21+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/i10xnews.kinsta.cloud\\\/harnessdev-llms-build-own-agent-harness\"},\"wordCount\":473,\"commentCount\":0,\"publisher\":{\"@id\":\"https:\\\/\\\/news.i10x.ai\\\/#organization\"},\"image\":{\"@id\":\"https:\\\/\\\/i10xnews.kinsta.cloud\\\/harnessdev-llms-build-own-agent-harness#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/i10x.ai\\\/news\\\/wp-content\\\/uploads\\\/2026\\\/10\\\/harnessdev-llms-build-own-agent-harness-title.webp\",\"articleSection\":[\"Agent Research\",\"Studies\"],\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"CommentAction\",\"name\":\"Comment\",\"target\":[\"https:\\\/\\\/i10xnews.kinsta.cloud\\\/harnessdev-llms-build-own-agent-harness#respond\"]}]},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/i10xnews.kinsta.cloud\\\/harnessdev-llms-build-own-agent-harness\",\"url\":\"https:\\\/\\\/i10xnews.kinsta.cloud\\\/harnessdev-llms-build-own-agent-harness\",\"name\":\"HarnessDev: Can LLMs Create Their Own Agent Harness? - i10X News\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/news.i10x.ai\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/i10xnews.kinsta.cloud\\\/harnessdev-llms-build-own-agent-harness#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/i10xnews.kinsta.cloud\\\/harnessdev-llms-build-own-agent-harness#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/i10x.ai\\\/news\\\/wp-content\\\/uploads\\\/2026\\\/10\\\/harnessdev-llms-build-own-agent-harness-title.webp\",\"datePublished\":\"2026-10-19T16:49:02+00:00\",\"dateModified\":\"2026-10-19T17:17:21+00:00\",\"description\":\"HarnessDev benchmark: LLM-built harnesses trail human ones on code\\\/search, match on writing\\\/ML. Evolution gains +3.11 points held-out.\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/i10xnews.kinsta.cloud\\\/harnessdev-llms-build-own-agent-harness#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/i10xnews.kinsta.cloud\\\/harnessdev-llms-build-own-agent-harness\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/i10xnews.kinsta.cloud\\\/harnessdev-llms-build-own-agent-harness#primaryimage\",\"url\":\"https:\\\/\\\/i10x.ai\\\/news\\\/wp-content\\\/uploads\\\/2026\\\/10\\\/harnessdev-llms-build-own-agent-harness-title.webp\",\"contentUrl\":\"https:\\\/\\\/i10x.ai\\\/news\\\/wp-content\\\/uploads\\\/2026\\\/10\\\/harnessdev-llms-build-own-agent-harness-title.webp\",\"width\":1344,\"height\":768,\"caption\":\"Illustration: i10X\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/i10xnews.kinsta.cloud\\\/harnessdev-llms-build-own-agent-harness#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/news.i10x.ai\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"LLMs Can Build Their Own Agent Harnesses \u2014 But Still Trail Human Ones on Code\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/news.i10x.ai\\\/#website\",\"url\":\"https:\\\/\\\/news.i10x.ai\\\/\",\"name\":\"i10X News\",\"description\":\"Stay updated with the latest news and announcements from i10X.\",\"publisher\":{\"@id\":\"https:\\\/\\\/news.i10x.ai\\\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/news.i10x.ai\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-US\"},{\"@type\":\"Organization\",\"@id\":\"https:\\\/\\\/news.i10x.ai\\\/#organization\",\"name\":\"i10X\",\"url\":\"https:\\\/\\\/news.i10x.ai\\\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/news.i10x.ai\\\/#\\\/schema\\\/logo\\\/image\\\/\",\"url\":\"https:\\\/\\\/i10xnews.kinsta.cloud\\\/wp-content\\\/uploads\\\/2026\\\/10\\\/i10x-logo-180.png\",\"contentUrl\":\"https:\\\/\\\/i10xnews.kinsta.cloud\\\/wp-content\\\/uploads\\\/2026\\\/10\\\/i10x-logo-180.png\",\"width\":180,\"height\":180,\"caption\":\"i10X\"},\"image\":{\"@id\":\"https:\\\/\\\/news.i10x.ai\\\/#\\\/schema\\\/logo\\\/image\\\/\"},\"sameAs\":[\"https:\\\/\\\/x.com\\\/i10x_ai\",\"https:\\\/\\\/www.linkedin.com\\\/company\\\/i10x-ai\",\"https:\\\/\\\/instagram.com\\\/i10x.ai\",\"https:\\\/\\\/www.youtube.com\\\/@i10X_ai_official\"]},{\"@type\":\"Person\",\"@id\":\"https:\\\/\\\/news.i10x.ai\\\/#\\\/schema\\\/person\\\/d410bb7ee72febefca2c9c369939219a\",\"name\":\"Christopher Ort\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/wp-content\\\/uploads\\\/2026\\\/10\\\/christopher-ort-192.webp\",\"url\":\"https:\\\/\\\/i10x.ai\\\/news\\\/wp-content\\\/uploads\\\/2026\\\/10\\\/christopher-ort-192.webp\",\"contentUrl\":\"https:\\\/\\\/i10x.ai\\\/news\\\/wp-content\\\/uploads\\\/2026\\\/10\\\/christopher-ort-192.webp\",\"caption\":\"Christopher Ort\"},\"description\":\"Christopher Ort handles SEO, AI agent workflows and PR & partnerships at i10x.ai, the all-in-one AI workspace built by i10X Global Pte. Ltd. in Singapore. He is a graduate of Friedrich-Alexander-Universit\u00e4t Erlangen-N\u00fcrnberg (FAU). He is the author of \u201cThe Style Penalty\u201d, a study published on SSRN that analyses 1,576 evaluation decisions by four frontier AI models to show how AI-written CVs fare in automated hiring screens; the research has been covered by heise online and t3n.\",\"url\":\"https:\\\/\\\/i10x.ai\\\/news\\\/author\\\/christopher-ort\",\"sameAs\":[\"https:\\\/\\\/i10x.ai\\\/blog\\\/author\\\/christopher-ort\"],\"worksFor\":{\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/#organization\"},\"alumniOf\":{\"@type\":\"EducationalOrganization\",\"name\":\"Friedrich-Alexander-Universit\u00e4t Erlangen-N\u00fcrnberg (FAU)\",\"url\":\"https:\\\/\\\/www.fau.eu\"}}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"HarnessDev: Can LLMs Create Their Own Agent Harness? - i10X News","description":"HarnessDev benchmark: LLM-built harnesses trail human ones on code\/search, match on writing\/ML. Evolution gains +3.11 points held-out.","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/i10xnews.kinsta.cloud\/harnessdev-llms-build-own-agent-harness","og_locale":"en_US","og_type":"article","og_title":"HarnessDev: Can LLMs Create Their Own Agent Harness? - i10X News","og_description":"HarnessDev benchmark: LLM-built harnesses trail human ones on code\/search, match on writing\/ML. Evolution gains +3.11 points held-out.","og_url":"https:\/\/i10xnews.kinsta.cloud\/harnessdev-llms-build-own-agent-harness","og_site_name":"i10X News","article_published_time":"2026-10-19T16:49:02+00:00","article_modified_time":"2026-10-19T17:17:21+00:00","og_image":[{"width":1344,"height":768,"url":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/harnessdev-llms-build-own-agent-harness-title.webp","type":"image\/webp"}],"author":"Christopher Ort","twitter_card":"summary_large_image","twitter_creator":"@i10x_ai","twitter_site":"@i10x_ai","twitter_misc":{"Written by":"Christopher Ort","Est. reading time":"2 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"NewsArticle","@id":"https:\/\/i10xnews.kinsta.cloud\/harnessdev-llms-build-own-agent-harness#article","isPartOf":{"@id":"https:\/\/i10xnews.kinsta.cloud\/harnessdev-llms-build-own-agent-harness"},"author":{"name":"Christopher Ort","@id":"https:\/\/news.i10x.ai\/#\/schema\/person\/d410bb7ee72febefca2c9c369939219a"},"headline":"LLMs Can Build Their Own Agent Harnesses \u2014 But Still Trail Human Ones on Code","datePublished":"2026-10-19T16:49:02+00:00","dateModified":"2026-10-19T17:17:21+00:00","mainEntityOfPage":{"@id":"https:\/\/i10xnews.kinsta.cloud\/harnessdev-llms-build-own-agent-harness"},"wordCount":473,"commentCount":0,"publisher":{"@id":"https:\/\/news.i10x.ai\/#organization"},"image":{"@id":"https:\/\/i10xnews.kinsta.cloud\/harnessdev-llms-build-own-agent-harness#primaryimage"},"thumbnailUrl":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/harnessdev-llms-build-own-agent-harness-title.webp","articleSection":["Agent Research","Studies"],"inLanguage":"en-US","potentialAction":[{"@type":"CommentAction","name":"Comment","target":["https:\/\/i10xnews.kinsta.cloud\/harnessdev-llms-build-own-agent-harness#respond"]}]},{"@type":"WebPage","@id":"https:\/\/i10xnews.kinsta.cloud\/harnessdev-llms-build-own-agent-harness","url":"https:\/\/i10xnews.kinsta.cloud\/harnessdev-llms-build-own-agent-harness","name":"HarnessDev: Can LLMs Create Their Own Agent Harness? - i10X News","isPartOf":{"@id":"https:\/\/news.i10x.ai\/#website"},"primaryImageOfPage":{"@id":"https:\/\/i10xnews.kinsta.cloud\/harnessdev-llms-build-own-agent-harness#primaryimage"},"image":{"@id":"https:\/\/i10xnews.kinsta.cloud\/harnessdev-llms-build-own-agent-harness#primaryimage"},"thumbnailUrl":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/harnessdev-llms-build-own-agent-harness-title.webp","datePublished":"2026-10-19T16:49:02+00:00","dateModified":"2026-10-19T17:17:21+00:00","description":"HarnessDev benchmark: LLM-built harnesses trail human ones on code\/search, match on writing\/ML. Evolution gains +3.11 points held-out.","breadcrumb":{"@id":"https:\/\/i10xnews.kinsta.cloud\/harnessdev-llms-build-own-agent-harness#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/i10xnews.kinsta.cloud\/harnessdev-llms-build-own-agent-harness"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/i10xnews.kinsta.cloud\/harnessdev-llms-build-own-agent-harness#primaryimage","url":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/harnessdev-llms-build-own-agent-harness-title.webp","contentUrl":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/harnessdev-llms-build-own-agent-harness-title.webp","width":1344,"height":768,"caption":"Illustration: i10X"},{"@type":"BreadcrumbList","@id":"https:\/\/i10xnews.kinsta.cloud\/harnessdev-llms-build-own-agent-harness#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/news.i10x.ai\/"},{"@type":"ListItem","position":2,"name":"LLMs Can Build Their Own Agent Harnesses \u2014 But Still Trail Human Ones on Code"}]},{"@type":"WebSite","@id":"https:\/\/news.i10x.ai\/#website","url":"https:\/\/news.i10x.ai\/","name":"i10X News","description":"Stay updated with the latest news and announcements from i10X.","publisher":{"@id":"https:\/\/news.i10x.ai\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/news.i10x.ai\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"},{"@type":"Organization","@id":"https:\/\/news.i10x.ai\/#organization","name":"i10X","url":"https:\/\/news.i10x.ai\/","logo":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/news.i10x.ai\/#\/schema\/logo\/image\/","url":"https:\/\/i10xnews.kinsta.cloud\/wp-content\/uploads\/2026\/10\/i10x-logo-180.png","contentUrl":"https:\/\/i10xnews.kinsta.cloud\/wp-content\/uploads\/2026\/10\/i10x-logo-180.png","width":180,"height":180,"caption":"i10X"},"image":{"@id":"https:\/\/news.i10x.ai\/#\/schema\/logo\/image\/"},"sameAs":["https:\/\/x.com\/i10x_ai","https:\/\/www.linkedin.com\/company\/i10x-ai","https:\/\/instagram.com\/i10x.ai","https:\/\/www.youtube.com\/@i10X_ai_official"]},{"@type":"Person","@id":"https:\/\/news.i10x.ai\/#\/schema\/person\/d410bb7ee72febefca2c9c369939219a","name":"Christopher Ort","image":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/christopher-ort-192.webp","url":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/christopher-ort-192.webp","contentUrl":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/christopher-ort-192.webp","caption":"Christopher Ort"},"description":"Christopher Ort handles SEO, AI agent workflows and PR & partnerships at i10x.ai, the all-in-one AI workspace built by i10X Global Pte. Ltd. in Singapore. He is a graduate of Friedrich-Alexander-Universit\u00e4t Erlangen-N\u00fcrnberg (FAU). He is the author of \u201cThe Style Penalty\u201d, a study published on SSRN that analyses 1,576 evaluation decisions by four frontier AI models to show how AI-written CVs fare in automated hiring screens; the research has been covered by heise online and t3n.","url":"https:\/\/i10x.ai\/news\/author\/christopher-ort","sameAs":["https:\/\/i10x.ai\/blog\/author\/christopher-ort"],"worksFor":{"@id":"https:\/\/i10x.ai\/news\/#organization"},"alumniOf":{"@type":"EducationalOrganization","name":"Friedrich-Alexander-Universit\u00e4t Erlangen-N\u00fcrnberg (FAU)","url":"https:\/\/www.fau.eu"}}]}},"_links":{"self":[{"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/posts\/6118","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/comments?post=6118"}],"version-history":[{"count":2,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/posts\/6118\/revisions"}],"predecessor-version":[{"id":6269,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/posts\/6118\/revisions\/6269"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/media\/6268"}],"wp:attachment":[{"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/media?parent=6118"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/categories?post=6118"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/tags?post=6118"},{"taxonomy":"company","embeddable":true,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/companies?post=6118"},{"taxonomy":"product","embeddable":true,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/products?post=6118"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}