Researchers at HKUST developed Hierarchical Self-Improvement (HSI), a framework enabling a single, frozen LLM to autonomously evolve its task-specific operational harness and the strategy for its own evolution. The system achieved substantial performance increases, such as +39.3% on BabyAI and +33.0% on Crafter, showing held-out generalization on navigation tasks while identifying feedback fidelity and backbone capability as key limitations.</p>\n","updatedAt":"2026-08-21T13:35:06.858Z","author":{"_id":"6938d0dc052bf4a9e3825af5","avatarUrl":"/avatars/b29a2715cc081e86bbf89a3ab238edab.svg","fullname":"Tailin","name":"TalentZHOU","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.9087753295898438},"editors":["TalentZHOU"],"editorAvatarUrls":["/avatars/b29a2715cc081e86bbf89a3ab238edab.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2608.08466","authors":[{"_id":"6a83b04e675db694db8cd43d","user":{"_id":"6938d0dc052bf4a9e3825af5","avatarUrl":"/avatars/b29a2715cc081e86bbf89a3ab238edab.svg","isPro":false,"fullname":"Tailin","user":"TalentZHOU","type":"user","name":"TalentZHOU"},"name":"Tailin Zhou","status":"claimed_verified","statusLastChangedAt":"2026-08-21T13:18:01.478Z","hidden":false}],"publishedAt":"2026-08-09T00:00:00.000Z","submittedOnDailyAt":"2026-08-21T00:00:00.000Z","title":"Hierarchical Self-Improvement: A Framework for Task-Specific Evolvable Agent Harnesses","submittedOnDailyBy":{"_id":"6938d0dc052bf4a9e3825af5","avatarUrl":"/avatars/b29a2715cc081e86bbf89a3ab238edab.svg","isPro":false,"fullname":"Tailin","user":"TalentZHOU","type":"user","name":"TalentZHOU"},"summary":"Modern LLM agents are often improved by modifying prompts, tools, or workflows manually, while the executable scaffold surrounding the model---the harness---is typically treated as a fixed artifact after deployment. This work studies an alternative where the harness is task-specific and continuously evolvable: each task family maintains its own harness, which is hot-swapped across iterations through a fixed task-injection seam and rewritten using environment feedback. We introduce Hierarchical Self-Improvement (HSI), a framework in which a single frozen LLM M operates across three hierarchical scopes: a task harness H that executes tasks, an evolver that rewrites H, and a meta-evolver that rewrites the evolver's strategy code under a frozen outer anchor. A thinking-on/off design isolates the contribution of harness evolution by disabling reasoning during task execution while enabling it during self-modification. HSI is bounded by two factors: a feedback-fidelity bound, since evolution requires informative reward signals to guide selection, and a backbone capability bound, since harness redesign cannot overcome limitations of the frozen model. On BALROG with DeepSeek-V4-Flash-Preview as the frozen backbone, HSI achieves consistent gains over the initial harness on moderate-difficulty tasks (+39.3 on BabyAI, +33.0 on Crafter, +25.0 on TextWorld, and +15.0 on MiniHack, all in raw \\% Progress), while obtaining strong held-out generalization on BabaIsAI sub-suites (0.98 best-test on BreakStop and 1.00 on GoTo from a 20% unseen split). On tasks beyond the backbone's capability (NLE), harness evolution provides no improvement. These results demonstrate task-specific harness evolution as a viable axis for improving frozen LLM agents under clear empirical limits. Code is available at https://github.com/TailinZhou/hsi.","upvotes":2,"discussionId":"6a83b04f675db694db8cd43e","githubRepo":"https://github.com/TailinZhou/hsi","githubRepoAddedBy":"user","ai_summary":"Hierarchical Self-Improvement evolves task-specific execution harnesses for frozen LLM agents via hierarchical self-modification, yielding substantial gains on moderate tasks while being bounded by feedback quality and backbone limits.","ai_keywords":["Hierarchical Self-Improvement","harness evolution","frozen LLM","meta-evolver","feedback-fidelity bound","backbone capability bound","task-injection seam"],"ai_summary_model":"thinkingmachines/Inkling-Small","githubStars":7,"organization":{"_id":"63355133edc1a61aecf74b0e","name":"HKUST","fullname":"HKUST","avatar":"https://www.gravatar.com/avatar/4a4318de793d2c187cb6f312e9d0e7bc?d=retro&size=100"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"6938d0dc052bf4a9e3825af5","avatarUrl":"/avatars/b29a2715cc081e86bbf89a3ab238edab.svg","isPro":false,"fullname":"Tailin","user":"TalentZHOU","type":"user"},{"_id":"63ac5701c21e60a3e9b58aa7","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/63ac5701c21e60a3e9b58aa7/g6EX7diOpuA94R2ab-rZC.png","isPro":true,"fullname":"Dipankar Sarkar","user":"dipankarsarkar","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"organization":{"_id":"63355133edc1a61aecf74b0e","name":"HKUST","fullname":"HKUST","avatar":"https://www.gravatar.com/avatar/4a4318de793d2c187cb6f312e9d0e7bc?d=retro&size=100"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2608/2608.08466.md","query":{}}">

Get this paper in your agent:

Don't have the latest CLI?

Cite arxiv.org/abs/2608.08466 in a model README.md to link it from this page.

Cite arxiv.org/abs/2608.08466 in a dataset README.md to link it from this page.

Cite arxiv.org/abs/2608.08466 in a Space README.md to link it from this page.

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