Key idea: Skills are not universally effective across LLM backbones. This paper introduces MASA, a model-aware skill alignment framework that rewrites and adapts skills based on a model’s capability profile. Results across multiple agent environments show that model-specific skill alignment significantly outperforms generic skill libraries and generalizes to new tasks with low inference cost.</p>\n","updatedAt":"2026-06-02T02:49:28.965Z","author":{"_id":"64d09c16c0c627dfa7f22599","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64d09c16c0c627dfa7f22599/TCV-PmAmPcbRpd2Nc11CL.jpeg","fullname":"jianxiangyu","name":"ffjasonyu","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":3,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.8690945506095886},"editors":["ffjasonyu"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/64d09c16c0c627dfa7f22599/TCV-PmAmPcbRpd2Nc11CL.jpeg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2605.30723","authors":[{"_id":"6a1d84fe808ddbc3c7d438c6","name":"Jianxiang Yu","hidden":false},{"_id":"6a1d84fe808ddbc3c7d438c7","name":"Jiapeng Zhu","hidden":false},{"_id":"6a1d84fe808ddbc3c7d438c8","name":"Bochen Lin","hidden":false},{"_id":"6a1d84fe808ddbc3c7d438c9","name":"Qier Cui","hidden":false},{"_id":"6a1d84fe808ddbc3c7d438ca","name":"Zichen Ding","hidden":false},{"_id":"6a1d84fe808ddbc3c7d438cb","name":"Xiang Li","hidden":false}],"publishedAt":"2026-05-29T00:00:00.000Z","submittedOnDailyAt":"2026-06-02T00:00:00.000Z","title":"Skill is Not One-Size-Fits-All: Model-Aware Skill Alignment for LLM Agents","submittedOnDailyBy":{"_id":"64d09c16c0c627dfa7f22599","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64d09c16c0c627dfa7f22599/TCV-PmAmPcbRpd2Nc11CL.jpeg","isPro":false,"fullname":"jianxiangyu","user":"ffjasonyu","type":"user","name":"ffjasonyu"},"summary":"LLM agents increasingly retrieve externally curated skills-procedural instructions retrieved at decision time-to improve performance on long-horizon interactive tasks. Existing skill libraries are typically treated as model-agnostic, reusing the same skill formulations across backbones with substantially different capacities and behaviors. However, our controlled experiments across multiple model scales show that skill effectiveness is strongly model-dependent: a skill that benefits one backbone can harm another. Motivated by this observation, we propose MASA Model-Aware Skill Alignment, a framework that adapts skills to each target backbone without modifying agent weights. MASA operates in two stages: (1) a hierarchical skill evolution pipeline that iteratively rewrites general and task-specific skills using hill climbing and UCB-driven tree search, guided by environment feedback and model capability profiles; and (2) a lightweight model-conditioned skill rewriter trained on evolution trajectories to reproduce the adaptation in a single forward pass. Experiments across three interactive environments and four backbones show that MASA consistently achieves the best overall performance, with gains of up to 25.8 points over the strongest baseline. The learned rewriter further generalizes to unseen tasks and environments without additional search, consistently outperforming a much larger teacher LLM at a fraction of the inference cost.","upvotes":7,"discussionId":"6a1d84ff808ddbc3c7d438cc","ai_summary":"Model-aware skill alignment framework adapts skills to different backbones through hierarchical evolution and lightweight rewriter training, achieving superior performance across interactive tasks.","ai_keywords":["LLM agents","skill libraries","model-agnostic","backbones","skill effectiveness","MASA","hierarchical skill evolution","hill climbing","UCB-driven tree search","model-conditioned skill rewriter","environment feedback","inference cost"]},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"620783f24e28382272337ba4","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/620783f24e28382272337ba4/zkUveQPNiDfYjgGhuFErj.jpeg","isPro":false,"fullname":"GuoLiangTang","user":"Tommy930","type":"user"},{"_id":"69ccceb5e3d65dcc2e45a495","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/IJ9PfOevsfEn6M0YQLw5h.png","isPro":false,"fullname":"한 서연","user":"emily-young450","type":"user"},{"_id":"66d8512c54209e9101811e8e","avatarUrl":"/avatars/62dfd8e6261108f2508efe678d5a2a57.svg","isPro":false,"fullname":"M Saad Salman","user":"MSS444","type":"user"},{"_id":"64d09c16c0c627dfa7f22599","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64d09c16c0c627dfa7f22599/TCV-PmAmPcbRpd2Nc11CL.jpeg","isPro":false,"fullname":"jianxiangyu","user":"ffjasonyu","type":"user"},{"_id":"698f8de8ae185b257313a76c","avatarUrl":"/avatars/488cad3f610f749a5631371fb2f019b8.svg","isPro":false,"fullname":"V9y2j9w0e","user":"v9y2j9w0e","type":"user"},{"_id":"63ca8e060609f1def7e6548a","avatarUrl":"/avatars/1da7947840cb87d5f77c0af9ee11f9c2.svg","isPro":true,"fullname":"Yi Jung","user":"YJ-142150","type":"user"},{"_id":"687363d49a81c7dcbcfa2d84","avatarUrl":"/avatars/5d943a5c811ed931c3fdcfee19253049.svg","isPro":false,"fullname":"jj","user":"realman123","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2605/2605.30723.md"}">

Get this paper in your agent:

Don't have the latest CLI?

Cite arxiv.org/abs/2605.30723 in a model README.md to link it from this page.

Cite arxiv.org/abs/2605.30723 in a dataset README.md to link it from this page.

Cite arxiv.org/abs/2605.30723 in a Space README.md to link it from this page.

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