@inproceedings{zhang-etal-2026-grounding,
title = "Grounding {LLM} Agent Intent in Multi-Module Environments with Executable Action Programs",
author = "Zhang, Jun and
Sun, Boyuan and
Piao, Jinghua and
Huang, Haoyu and
Jin, Depeng and
Li, Yong",
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.2226/",
pages = "50666--50685",
ISBN = "979-8-89176-459-0",
abstract = "Large Language Model (LLM) agents are increasingly expected to interact with external environments rather than only generate text. A key challenge in this process is to accurately and efficiently ground high-level agent intents into valid executable action sequences, especially when the environment consists of multiple interacting modules. To address this challenge, we introduce \textit{executable action programs} (EAPs), a restricted Python-based intermediate representation between textual agent intents and structured environment-module action calls, which leverages the expressive power of Python to explicitly represent complex behavioral sequences, their branching relationships, and cross-module data dependencies. Based on this representation, we implement \textbf{EAPRouter}, which generates, validates, executes, and reuses EAPs through an agent-EAP interface and a lightweight environment module integration protocol. We evaluate EAPRouter on real agent-environment interaction traces from social simulation, a demanding application scenario with multi-agent, multi-module, and repeated interaction patterns. Experiments show that EAPRouter achieves a 60.0{\%} grounding success rate with Kimi-K2-Instruct, outperforming the SOTA function-calling baseline by 31.6{\%}. It also improves over the strongest baselines by 11.5{\%}-25.4{\%} on GLM-4.7, Claude Opus 4.5, and Gemini 2.5 Flash, with smaller gains on less code-capable models. On the cross-domain NESTFUL v2 benchmark, EAPRouter matches ReAct{'}s executable accuracy (83.8{\%} vs. 84.9{\%}) while reducing LLM calls, tokens, and execution time by about 80{\%}. EAP pre-generation further handles over 70{\%} of interaction requests, cache reuse covers 66.5{\%} of the remaining ones, and safety validation blocks all tested unsafe payloads and prompt-injection attempts. The code is available at https://github.com/tsinghua-fib-lab/EAPRouter."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="zhang-etal-2026-grounding">
<titleInfo>
<title>Grounding LLM Agent Intent in Multi-Module Environments with Executable Action Programs</title>
</titleInfo>
<name type="personal">
<namePart type="given">Jun</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Boyuan</namePart>
<namePart type="family">Sun</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jinghua</namePart>
<namePart type="family">Piao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Haoyu</namePart>
<namePart type="family">Huang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Depeng</namePart>
<namePart type="family">Jin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yong</namePart>
<namePart type="family">Li</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>Large Language Model (LLM) agents are increasingly expected to interact with external environments rather than only generate text. A key challenge in this process is to accurately and efficiently ground high-level agent intents into valid executable action sequences, especially when the environment consists of multiple interacting modules. To address this challenge, we introduce executable action programs (EAPs), a restricted Python-based intermediate representation between textual agent intents and structured environment-module action calls, which leverages the expressive power of Python to explicitly represent complex behavioral sequences, their branching relationships, and cross-module data dependencies. Based on this representation, we implement EAPRouter, which generates, validates, executes, and reuses EAPs through an agent-EAP interface and a lightweight environment module integration protocol. We evaluate EAPRouter on real agent-environment interaction traces from social simulation, a demanding application scenario with multi-agent, multi-module, and repeated interaction patterns. Experiments show that EAPRouter achieves a 60.0% grounding success rate with Kimi-K2-Instruct, outperforming the SOTA function-calling baseline by 31.6%. It also improves over the strongest baselines by 11.5%-25.4% on GLM-4.7, Claude Opus 4.5, and Gemini 2.5 Flash, with smaller gains on less code-capable models. On the cross-domain NESTFUL v2 benchmark, EAPRouter matches ReAct’s executable accuracy (83.8% vs. 84.9%) while reducing LLM calls, tokens, and execution time by about 80%. EAP pre-generation further handles over 70% of interaction requests, cache reuse covers 66.5% of the remaining ones, and safety validation blocks all tested unsafe payloads and prompt-injection attempts. The code is available at https://github.com/tsinghua-fib-lab/EAPRouter.</abstract>
<identifier type="citekey">zhang-etal-2026-grounding</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.2226/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>50666</start>
<end>50685</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Grounding LLM Agent Intent in Multi-Module Environments with Executable Action Programs
%A Zhang, Jun
%A Sun, Boyuan
%A Piao, Jinghua
%A Huang, Haoyu
%A Jin, Depeng
%A Li, Yong
%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 zhang-etal-2026-grounding
%X Large Language Model (LLM) agents are increasingly expected to interact with external environments rather than only generate text. A key challenge in this process is to accurately and efficiently ground high-level agent intents into valid executable action sequences, especially when the environment consists of multiple interacting modules. To address this challenge, we introduce executable action programs (EAPs), a restricted Python-based intermediate representation between textual agent intents and structured environment-module action calls, which leverages the expressive power of Python to explicitly represent complex behavioral sequences, their branching relationships, and cross-module data dependencies. Based on this representation, we implement EAPRouter, which generates, validates, executes, and reuses EAPs through an agent-EAP interface and a lightweight environment module integration protocol. We evaluate EAPRouter on real agent-environment interaction traces from social simulation, a demanding application scenario with multi-agent, multi-module, and repeated interaction patterns. Experiments show that EAPRouter achieves a 60.0% grounding success rate with Kimi-K2-Instruct, outperforming the SOTA function-calling baseline by 31.6%. It also improves over the strongest baselines by 11.5%-25.4% on GLM-4.7, Claude Opus 4.5, and Gemini 2.5 Flash, with smaller gains on less code-capable models. On the cross-domain NESTFUL v2 benchmark, EAPRouter matches ReAct’s executable accuracy (83.8% vs. 84.9%) while reducing LLM calls, tokens, and execution time by about 80%. EAP pre-generation further handles over 70% of interaction requests, cache reuse covers 66.5% of the remaining ones, and safety validation blocks all tested unsafe payloads and prompt-injection attempts. The code is available at https://github.com/tsinghua-fib-lab/EAPRouter.
%U https://aclanthology.org/2026.emnlp-main.2226/
%P 50666-50685
Markdown (Informal)
[Grounding LLM Agent Intent in Multi-Module Environments with Executable Action Programs](https://aclanthology.org/2026.emnlp-main.2226/) (Zhang et al., EMNLP 2026)
ACL