We propose PILOT, a supervisor–worker harness that makes agent self-improvement live rather than post-hoc. Two coupled mechanisms: a separate supervisor redirects or aborts the active worker mid-run (live steering), while runtime-discovered procedures and failure modes are distilled into reusable skills and memory (live self-evolution). With frozen GLM-5.1 and Kimi-K2.6 backbones, PILOT ranks first in 5 of 6 configurations across three benchmarks. In the self-improvement setting it gains +14.6 / +12.4 points while cutting mean output tokens by 42.9% / 47.4%, and successful evaluations per million tokens rise by 110.3% / 134.0%.</p>\n<p>Code will be released soon on GitHub.</p>\n","updatedAt":"2026-08-28T05:26:05.892Z","author":{"_id":"6002c316698168af3bb9f4a6","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6002c316698168af3bb9f4a6/M2J2QFCRc5RhYRoz7OuZN.png","fullname":"yangxiao","name":"YangXiao-nlp","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":7,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.8757508993148804},"editors":["YangXiao-nlp"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/6002c316698168af3bb9f4a6/M2J2QFCRc5RhYRoz7OuZN.png"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2608.26530","authors":[{"_id":"6a90f454a64059bab69c3609","user":{"_id":"6002c316698168af3bb9f4a6","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6002c316698168af3bb9f4a6/M2J2QFCRc5RhYRoz7OuZN.png","isPro":false,"fullname":"yangxiao","user":"YangXiao-nlp","type":"user","name":"YangXiao-nlp"},"name":"Yang Xiao","status":"claimed_verified","statusLastChangedAt":"2026-08-28T08:45:04.622Z","hidden":false},{"_id":"6a90f454a64059bab69c360a","name":"Yusong Sun","hidden":false},{"_id":"6a90f454a64059bab69c360b","name":"Haoyi Wu","hidden":false},{"_id":"6a90f454a64059bab69c360c","name":"Wenyang Hui","hidden":false},{"_id":"6a90f454a64059bab69c360d","name":"Wen Da","hidden":false},{"_id":"6a90f454a64059bab69c360e","name":"Zhaokai Luo","hidden":false},{"_id":"6a90f454a64059bab69c360f","name":"Mu Chuan","hidden":false},{"_id":"6a90f454a64059bab69c3610","name":"Yao Hu","hidden":false},{"_id":"6a90f454a64059bab69c3611","name":"Wenjie Li","hidden":false},{"_id":"6a90f454a64059bab69c3612","name":"Chengyue Jiang","hidden":false}],"publishedAt":"2026-08-27T00:00:00.000Z","submittedOnDailyAt":"2026-08-28T00:00:00.000Z","title":"PILOT in the Loop: Live Self-Improvement for Long-Horizon Agents","submittedOnDailyBy":{"_id":"6002c316698168af3bb9f4a6","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6002c316698168af3bb9f4a6/M2J2QFCRc5RhYRoz7OuZN.png","isPro":false,"fullname":"yangxiao","user":"YangXiao-nlp","type":"user","name":"YangXiao-nlp"},"summary":"Long-horizon agent runs generate experience that can improve both the current run and future work. Most self-improvement methods process this experience only after execution ends, so they cannot redirect the active run or immediately apply and validate lessons learned from it. We argue that self-improvement should instead be live, using emerging experience both to redirect the active run and to update the persistent harness. Existing agent architectures do not fully support this goal. Single-agent self-correction combines task execution and trajectory assessment within one context, while subagent delegation separates execution but typically cannot redirect an active subagent. We present PILOT, a supervisor-worker harness for live self-improvement through two coupled mechanisms: (1) live steering lets a separate supervisor redirect or abort the active worker during execution; and (2) live self-evolution distils procedures and failure modes revealed during execution into reusable skills and memory. Across two frozen backbones and three benchmarks, PILOT ranks first in five of six configurations. On Terminal-Bench 2.0, PILOT outperforms counterpart harnesses by up to 9.8 percentage points. In the self-improvement setting, PILOT gains 14.6 points with GLM-5.1 and 12.4 points with Kimi-K2.6. Mean output tokens fall by 42.9% and 47.4%, while successful evaluations per million output tokens rise by 110.3% and 134.0%, respectively.","upvotes":21,"discussionId":"6a90f454a64059bab69c3613","ai_summary":"PILOT enables live self-improvement by allowing a supervisor to steer active workers and distilling execution experience into reusable skills, improving accuracy and efficiency.","ai_keywords":["live steering","supervisor-worker harness","live self-evolution","reusable skills","memory","self-correction","subagent delegation"],"ai_summary_model":"thinkingmachines/Inkling-Small","organization":{"_id":"646ecc368d316fde87b3b6e3","name":"PolyUHK","fullname":"The Hong Kong Polytechnic University","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/646ecbc0cbb7bb996513e298/Akb4zKqIP9kb9PQoUPUmj.jpeg"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"6002c316698168af3bb9f4a6","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6002c316698168af3bb9f4a6/M2J2QFCRc5RhYRoz7OuZN.png","isPro":false,"fullname":"yangxiao","user":"YangXiao-nlp","type":"user"},{"_id":"69537e9c313d96feb976d0d1","avatarUrl":"/avatars/80a8295a67d4d7dfec80ab8359bd5612.svg","isPro":false,"fullname":"Yusong Sun","user":"YusongSun","type":"user"},{"_id":"64fd753e5ca946a010d61b54","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/HbbwJZqxghoR-57v3Bq8l.jpeg","isPro":false,"fullname":"Hui Zhang","user":"Julie11","type":"user"},{"_id":"6596534fd61d4cbc1bf08683","avatarUrl":"/avatars/ab37f077bcebccb52a1eb2b0b7567552.svg","isPro":false,"fullname":"Jiashuo WANG","user":"Jessie09","type":"user"},{"_id":"64e6c617ecce34cb442cb208","avatarUrl":"/avatars/ebc61bf6a043314cb2089b1efd5e6a18.svg","isPro":false,"fullname":"JieSun(SII)","user":"Sunshine279","type":"user"},{"_id":"6a911cdb8f844ed4db749a6a","avatarUrl":"/avatars/877f0a6b62535b8cb15f0af9a2c776ab.svg","isPro":false,"fullname":"Jeff","user":"Jeff0628","type":"user"},{"_id":"671b8777ac4168f79848b282","avatarUrl":"/avatars/b4b8062a8fe890fb6bac8917630bfb5a.svg","isPro":false,"fullname":"Zechen Sun","user":"Mintszc","type":"user"},{"_id":"6a911df0dd062b16f11e400e","avatarUrl":"/avatars/da2dc0be8747cc4f9f69b0e42330f995.svg","isPro":false,"fullname":"huiwy","user":"huiwy1","type":"user"},{"_id":"6469f91531fbfc5df81b9154","avatarUrl":"/avatars/20658b7341f181dbe702b4665d08780d.svg","isPro":false,"fullname":"whywhy","user":"whynlp","type":"user"},{"_id":"6a911ff3c72880ee4efb22e1","avatarUrl":"/avatars/7405273092bd756a0f8584c363414382.svg","isPro":false,"fullname":"Reiko","user":"Reiko777","type":"user"},{"_id":"69a6944d0a20e2f2f984aaee","avatarUrl":"/avatars/1c60bf82930125e2cb7afe0b94ca4288.svg","isPro":false,"fullname":"yuaofan","user":"AofaYu71","type":"user"},{"_id":"630341a4ef6f432d84e97db9","avatarUrl":"/avatars/dc9910b5b9053123bad5ff848f677e12.svg","isPro":false,"fullname":"tomsawyer","user":"tomhu","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"organization":{"_id":"646ecc368d316fde87b3b6e3","name":"PolyUHK","fullname":"The Hong Kong Polytechnic University","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/646ecbc0cbb7bb996513e298/Akb4zKqIP9kb9PQoUPUmj.jpeg"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2608/2608.26530.md","query":{}}">

Get this paper in your agent:

Don't have the latest CLI?

Cite arxiv.org/abs/2608.26530 in a model README.md to link it from this page.

Cite arxiv.org/abs/2608.26530 in a dataset README.md to link it from this page.

Cite arxiv.org/abs/2608.26530 in a Space README.md to link it from this page.

Add this paper to a collection to link it from this page.