Hugging Face Daily Papers · August 31, 2026 · 4 min read

ContextPilot: Teaching Agents for Proactive Context Management via Fine-grained RL

Mirrored from Hugging Face Daily Papers for archival readability. Support the source by reading on the original site.

Like Read original ↗

ContextPilot extends context management with planning, structured memory, and soft context offloading. Its context-aware partial rollout focuses exploration on sensitive context-editing decisions, while fine-grained credit assignment trains intermediate snapshots using the outcomes of their downstream branches. Experiments on long-context QA and deep search tasks show that ContextPilot achieves stronger performance with a more compact working context, outperforming existing baselines across various base models and benchmarks.</p>\n","updatedAt":"2026-08-31T01:49:52.509Z","author":{"_id":"64ddbc3f1f2dad27e1a05ac1","avatarUrl":"/avatars/4fb2753b7998c8536bfd4780d3b10a6d.svg","fullname":"Junrulu","name":"Junrulu","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":15,"isUserFollowing":false,"primaryOrg":{"avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/Lp3m-XLpjQGwBItlvn69q.png","fullname":"Tencent","name":"tencent","type":"org","isHf":false,"plan":"team"}}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.8410558700561523},"editors":["Junrulu"],"editorAvatarUrls":["/avatars/4fb2753b7998c8536bfd4780d3b10a6d.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2608.28476","authors":[{"_id":"6a94da5e073195fee51571d8","name":"Zhuoshi Pan","hidden":false},{"_id":"6a94da5e073195fee51571d9","name":"Qizhi Pei","hidden":false},{"_id":"6a94da5e073195fee51571da","name":"Junru Lu","hidden":false},{"_id":"6a94da5e073195fee51571db","name":"Honglin Lin","hidden":false},{"_id":"6a94da5e073195fee51571dc","name":"H. Vicky Zhao","hidden":false},{"_id":"6a94da5e073195fee51571dd","name":"Di Yin","hidden":false},{"_id":"6a94da5e073195fee51571de","name":"Xing Sun","hidden":false}],"publishedAt":"2026-08-28T00:00:00.000Z","submittedOnDailyAt":"2026-08-31T00:00:00.000Z","title":"ContextPilot: Teaching Agents for Proactive Context Management via Fine-grained RL","submittedOnDailyBy":{"_id":"64ddbc3f1f2dad27e1a05ac1","avatarUrl":"/avatars/4fb2753b7998c8536bfd4780d3b10a6d.svg","isPro":false,"fullname":"Junrulu","user":"Junrulu","type":"user","name":"Junrulu"},"summary":"Long-horizon agentic tasks require large language models (LLMs) to iteratively retrieve, integrate, and maintain dispersed information across multi-turn interactions, but preserving all interaction histories leads to a continuously growing working context. Recent proactive context management methods allow models to edit their own working context with specialized tools, yet they still face three key limitations: (1) a limited toolset restricted to search, deletion, and summarization, with no support for global planning, long-term memory, and adaptive compression; (2) inefficient exploration that treats context management actions uniformly despite their heterogeneous impacts on final outcomes; and (3) coarse-grained credit assignment that assigns the final trajectory-level reward to all intermediate context editing actions during RL. To bridge these gaps, we introduce ContextPilot, a proactive context management framework for long-horizon agentic reasoning. Our approach systematically augments the toolset with planning, long-term memory, and soft context offloading tools. We further propose an RL method tailored for context management, which uses context and entropy variation to identify critical editing decisions for branch sampling and estimates action-level advantages from all branched trajectories that pass through the corresponding context editing action. Experiments on long-context QA and deep search tasks show that ContextPilot achieves stronger performance with a more compact working context, consistently outperforming existing baselines across various base models and benchmarks. Code is available at https://github.com/Tencent/ContextPilot.","upvotes":21,"discussionId":"6a94da5f073195fee51571df","projectPage":"https://tencent.github.io/ContextPilot","githubRepo":"https://github.com/Tencent/ContextPilot","githubRepoAddedBy":"user","ai_summary":"ContextPilot improves long-horizon agent reasoning by expanding context-editing tools and using reinforcement learning with branch sampling to identify critical context decisions.","ai_keywords":["proactive context management","long-horizon agentic reasoning","context editing tools","planning","long-term memory","soft context offloading","reinforcement learning","branch sampling","action-level advantages","entropy variation"],"ai_summary_model":"thinkingmachines/Inkling-Small","githubStars":2,"organization":{"_id":"66543b6e420092799d2f625c","name":"tencent","fullname":"Tencent","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/Lp3m-XLpjQGwBItlvn69q.png"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"64ddbc3f1f2dad27e1a05ac1","avatarUrl":"/avatars/4fb2753b7998c8536bfd4780d3b10a6d.svg","isPro":false,"fullname":"Junrulu","user":"Junrulu","type":"user"},{"_id":"6565e44b9bf6665f10016213","avatarUrl":"/avatars/0d7cf0e1b42cd116960cd030478446c5.svg","isPro":false,"fullname":"Zhuoshi Pan","user":"panzs19","type":"user"},{"_id":"640d99628512ec51d7ef71c7","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/640d99628512ec51d7ef71c7/fcBkqnxfxuuuZTqfN_BGy.jpeg","isPro":false,"fullname":"Honglin Lin","user":"LHL3341","type":"user"},{"_id":"6397f6081323f19c578f142e","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6397f6081323f19c578f142e/it7FYYKjlLX8wSsMLm8EO.jpeg","isPro":false,"fullname":"QizhiPei","user":"QizhiPei","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"},{"_id":"6363ad21287b5ce02ed187ae","avatarUrl":"/avatars/9903cddc7330852fc6d8cf672cebcb56.svg","isPro":false,"fullname":"SiyuAn","user":"moonlake","type":"user"},{"_id":"654c4f6d848e8c8b118e16e3","avatarUrl":"/avatars/b2323e95624e28c2578a3c5d73020c77.svg","isPro":false,"fullname":"Arnold Yang","user":"B3rrYang","type":"user"},{"_id":"63fc75f9b9db84750cea9c5c","avatarUrl":"/avatars/2c5bf9685e0cfc4b5785a4a86c34e0db.svg","isPro":false,"fullname":"DI YIN","user":"DIYIN","type":"user"},{"_id":"68761ea2602bdfd2d12940a6","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/MDJ4KnwufjX8_8CkZSOdW.png","isPro":false,"fullname":"richard_dxx","user":"dxxth0356","type":"user"},{"_id":"6a94ef826d49926c37ccaecf","avatarUrl":"/avatars/5d09fe269167b4eec6a36952672b47a0.svg","isPro":false,"fullname":"Ji Qi","user":"SingularityUndefined","type":"user"},{"_id":"61c93839af1a6e3b53ca6dbc","avatarUrl":"/avatars/7a68ff9d7fc0c2f83aaf5b541d90d9a8.svg","isPro":false,"fullname":"zefengdu","user":"seeledu","type":"user"},{"_id":"6a73dd247197e796b1b421d9","avatarUrl":"/avatars/5e7667ffaae4c007470c60c49d26ade1.svg","isPro":false,"fullname":"洪明","user":"sjalngh","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"organization":{"_id":"66543b6e420092799d2f625c","name":"tencent","fullname":"Tencent","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/Lp3m-XLpjQGwBItlvn69q.png"},"query":{}}">

Papers

arxiv:2608.28476

ContextPilot: Teaching Agents for Proactive Context Management via Fine-grained RL

Published on Aug 28

· Submitted by

Junrulu on Aug 31

Tencent

Upvote

Authors:

Abstract

ContextPilot improves long-horizon agent reasoning by expanding context-editing tools and using reinforcement learning with branch sampling to identify critical context decisions.

Generated by thinkingmachines/Inkling-Small

Long-horizon agentic tasks require large language models (LLMs) to iteratively retrieve, integrate, and maintain dispersed information across multi-turn interactions, but preserving all interaction histories leads to a continuously growing working context. Recent proactive context management methods allow models to edit their own working context with specialized tools, yet they still face three key limitations: (1) a limited toolset restricted to search, deletion, and summarization, with no support for global planning, long-term memory, and adaptive compression; (2) inefficient exploration that treats context management actions uniformly despite their heterogeneous impacts on final outcomes; and (3) coarse-grained credit assignment that assigns the final trajectory-level reward to all intermediate context editing actions during RL. To bridge these gaps, we introduce ContextPilot, a proactive context management framework for long-horizon agentic reasoning. Our approach systematically augments the toolset with planning, long-term memory, and soft context offloading tools. We further propose an RL method tailored for context management, which uses context and entropy variation to identify critical editing decisions for branch sampling and estimates action-level advantages from all branched trajectories that pass through the corresponding context editing action. Experiments on long-context QA and deep search tasks show that ContextPilot achieves stronger performance with a more compact working context, consistently outperforming existing baselines across various base models and benchmarks. Code is available at https://github.com/Tencent/ContextPilot.