@inproceedings{pan-etal-2026-contextpilot,
title = "{C}ontext{P}ilot: Teaching Agents for Proactive Context Management via Fine-grained {RL}",
author = "Pan, Zhuoshi and
Pei, Qizhi and
Lu, Junru and
Lin, Honglin and
Zhao, H. Vicky and
Yin, di and
Sun, Xing",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.299/",
pages = "6886--6903",
ISBN = "979-8-89176-459-0",
abstract = "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 context control actions for branch sampling and estimates action-level advantages from all branched trajectories that pass through these actions. 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."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="pan-etal-2026-contextpilot">
<titleInfo>
<title>ContextPilot: Teaching Agents for Proactive Context Management via Fine-grained RL</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zhuoshi</namePart>
<namePart type="family">Pan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Qizhi</namePart>
<namePart type="family">Pei</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Junru</namePart>
<namePart type="family">Lu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Honglin</namePart>
<namePart type="family">Lin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">H</namePart>
<namePart type="given">Vicky</namePart>
<namePart type="family">Zhao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">di</namePart>
<namePart type="family">Yin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xing</namePart>
<namePart type="family">Sun</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>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 context control actions for branch sampling and estimates action-level advantages from all branched trajectories that pass through these actions. 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.</abstract>
<identifier type="citekey">pan-etal-2026-contextpilot</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.299/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>6886</start>
<end>6903</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T ContextPilot: Teaching Agents for Proactive Context Management via Fine-grained RL
%A Pan, Zhuoshi
%A Pei, Qizhi
%A Lu, Junru
%A Lin, Honglin
%A Zhao, H. Vicky
%A Yin, di
%A Sun, Xing
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F pan-etal-2026-contextpilot
%X 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 context control actions for branch sampling and estimates action-level advantages from all branched trajectories that pass through these actions. 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.
%U https://aclanthology.org/2026.emnlp-main.299/
%P 6886-6903
Markdown (Informal)
[ContextPilot: Teaching Agents for Proactive Context Management via Fine-grained RL](https://aclanthology.org/2026.emnlp-main.299/) (Pan et al., EMNLP 2026)
ACL