{"id":52426,"date":"2026-04-24T14:15:30","date_gmt":"2026-04-24T06:15:30","guid":{"rendered":"https:\/\/www.1ai.net\/?p=52426"},"modified":"2026-04-24T14:15:30","modified_gmt":"2026-04-24T06:15:30","slug":"augment-code-%e5%ae%9e%e6%b5%8b%ef%bc%9a%e5%a5%bd%e7%9a%84-agents-md-%e7%9b%b8%e5%bd%93%e4%ba%8e%e6%a8%a1%e5%9e%8b%e5%8d%87%e7%ba%a7%e4%b8%80%e6%a1%a3%ef%bc%8c%e5%b7%ae%e7%9a%84%e4%b8%8d%e5%a6%82","status":"publish","type":"post","link":"https:\/\/www.1ai.net\/en\/52426.html","title":{"rendered":"Automation Code: Good AGENTS.md equals a model upgrade"},"content":{"rendered":"<p>APRIL 24TH, AI PROGRAMMING TOOL <a href=\"https:\/\/www.1ai.net\/en\/tag\/augment-code\" title=\"_Other Organiser\" target=\"_blank\" >August Code<\/a> Yesterday, an internal study was published: extract dozens of AGENTS.mds from their home code library, measuring their coding AI <a href=\"https:\/\/www.1ai.net\/en\/tag\/%e6%99%ba%e8%83%bd%e4%bd%93\" title=\"[View articles tagged with [intelligent body]]\" target=\"_blank\" >Agent<\/a>Actual impact of output quality\u3002<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-52427\" title=\"434d3d35j00tzioo001ud000u1g1m\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2026\/04\/434d3d35j00tdzioo001ud000u000g1m.jpg\" alt=\"434d3d35j00tzioo001ud000u1g1m\" width=\"1080\" height=\"577\" \/><\/p>\n<p>&nbsp;<\/p>\n<p><strong>The results show that a well-written AGENTS.md brings an upgrade equivalent to a model that went from Haiku to Opus, which is not written at all\u3002<\/strong><\/p>\n<p>The study found that the valid AGENTS.md should keep the main file at 100 to 150 rows, with a small number of focused reference documents, to bring a full quality improvement of 10% to 15% in a medium-sized module\u3002<\/p>\n<p>TWO OF THE MOST SIGNIFICANT APPROACHES ARE TO WRITE THE PROCESS INTO A NUMBERING STEP AND USE THE DECISION SHEET TO HELP THE AI SMART BODY LOCK THE PROGRAM IN ADVANCE; THE BAN MUST BE WRITTEN IN CONJUNCTION WITH AN ALTERNATIVE SOLUTION, OTHERWISE THE SMART BODY WILL BE CAUGHT IN THE TRAP OF RETRACING AND DELAYING\u3002<\/p>\n<p>The most common reason for failure is that there is too much documentation. Once the intelligence body is attracted to a large number of architecture documents, the output changes when the mass context is loaded\u3002<\/p>\n<p>It was also noted that AGENTS.md was the only document that an intelligent body, 100%, would automatically read, stored under \\\\docs\/catalogues and not quoted, and found less than 10%\u3002<\/p>","protected":false},"excerpt":{"rendered":"<p>On April 24th, the AI Program Tool Company Augment Code released an internal study yesterday: extracting dozens of AGENTS.md from its home code library to measure its actual impact on the quality of coding AI smarts output. The results show that a well-written AGENTS.md brings an upgrade equivalent to a model that went from Haiku to Opus, which is not written at all. The study found that the valid AGENTS.md should keep the main file at 100 to 150 rows, with a small number of focused reference documents, to bring a full quality improvement of 10% to 15% in a medium-sized module. Two of the most significant practices are the operation flow<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[146],"tags":[8479,1355],"collection":[],"class_list":["post-52426","post","type-post","status-publish","format-standard","hentry","category-news","tag-augment-code","tag-1355"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/52426","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/comments?post=52426"}],"version-history":[{"count":0,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/52426\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=52426"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=52426"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=52426"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=52426"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}