Paper page - Code as Agent Harness

Source: https://huggingface.co/papers/2605.18747 Authors:

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

Abstract

Large language models are increasingly used as operational substrates for agent reasoning and execution in agentic systems, with code serving as a unified infrastructure layer across multiple domains and applications.

Recentlarge language models(LLMs) have demonstrated strong capabilities in understanding and generating code, from competitive programming to repository-level software engineering. In emergingagentic systems, code is no longer only a target output. It increasingly serves as an operational substrate for agentreasoning, acting,environment modeling, and execution-based verification. We frame this shift through the lens ofagent harnesses and introducecode as agent harness: a unified view that centers code as the basis foragent infrastructure. To systematically study this perspective, we organize the survey around three connected layers. First, we study the harness interface, where code connects agents toreasoning,action, andenvironment modeling. Second, we examine harness mechanisms:planning,memory, andtool usefor long-horizon execution, together withfeedback-driven controlandoptimizationthat make harness reliable and adaptive. Third, we discuss scaling the harness from single-agent systems to multi-agent settings, where shared code artifacts supportmulti-agent coordination, review, and verification. Across these layers, we summarize representative methods and practical applications ofcode as agent harness, spanning coding assistants, GUI/OS automation, embodied agents, scientific discovery, personalization and recommendation,DevOps, andenterprise workflows. We further outline open challenges for harness engineering, including evaluation beyond final task success, verification under incomplete feedback, regression-free harness improvement, consistent shared state across multiple agents, human oversight for safety-criticalactions, and extensions to multimodal environments. By centering code as the harness of agentic AI, this survey provides a unified roadmap toward executable, verifiable, and stateful AI agent systems.

View arXiv pageView PDFGitHub17Add to collection

Get this paper in your agent:

hf papers read 2605\.18747

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2605.18747 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2605.18747 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2605.18747 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.