Skip to content

How Should Agents See Code? Rethinking Code Representations for Agentic Software Engineering

Nov 2026 · 1 citation · 36 references
Computer Science

TL;DR

The results show that map-based code-representation workflows are most useful as token-saving tools when localization is the bottleneck, while their success gains narrow when the decisive step is patch or test authoring.

Abstract

LLM agents are increasingly used for repository-level software engineering tasks. However, most agents still access repositories through raw files, lexical search, and shell commands, treating code primarily as text. This paper investigates map-based code-representation workflows as a design variable in agentic software engineering, examining how different repository views change agents'retrieval behavior, task success, token use, and latency. In this paper, we study five code-representation workflows: abstract syntax trees, call graphs, module dependencies, documentation-enriched search, and a hybrid workflow that combines them. Each workflow exposes a repository map that the agent can query during a task. We evaluate Claude Code with Haiku 4.5 against a raw-tool baseline on four repository-level tasks: fault localization with LinuxFLBench, repository-level code generation and repair with SWE-bench Pro, test generation with SWT-bench, and security patching with SecBench. The results show that representation workflows are most useful as token-saving tools when localization is the bottleneck, while their success gains narrow when the decisive step is patch or test authoring. On LinuxFLBench, all five workflows improve recall@10 over the baseline (0.757-0.829 vs. 0.717) and reduce token usage by 25-39%. On SWE-bench Pro, resolution remains close to the baseline (44.3-50.6% vs. 49.0%) while token usage drops by 17-26%. On SecBench, Callgraph has the highest patch-resolution rate in this sample (34.3% vs. 28.9%) and uses fewer tokens, but all workflows increase latency. On SWT-bench, Hybrid and Docs have the highest reproducing-test rates in this sample (47.8% and 47.2% vs. 44.7%), but all workflows use more tokens and take longer. Overall, map-based workflows change both what the agent inspects and how much it spends, but their benefits are task-dependent rather than universal.

View source

Similar papers

Book Open access Oct 2026

Agent-HAnS: Maintaining Feature and Code Traceability in Agentic Development

This work presents Agent-HAnS, an agentic workflow which instructs the LLM to update a feature model and generate embedded feature annotations after each code-generation step, and visualizes the evolving structure on a live dashboard served via a Model Context Protocol server.

Johan Martinson, Jacob Bengtsson, Moltas Hultin et al. · 0 citations
#artificial intelligence Preprint Oct 2026

Evaluating Local Language Model Agents for Reproducible Data Engineering: An Empirical Software Engineering Study of Mobility Workflows

Context: Large language model (LLM) agents are increasingly used as software and data-engineering assistants, yet evidence about locally deployable open-weight agents remains limited. Existing evaluations often emphasize textual responses or isolated code generation rather than the validity of complete engineering arti...

Jorge García-Carrasco, Javier Sanchis, A. Reina-Reina et al. · 0 citations

E2E-SWE: Benchmarking LLMs on Building Working Codebases from Scratch

E2E-SWE is introduced, a benchmark for evaluating whether coding agents can build complete, functional software repositories end to end and finds substantial variation in end-to-end repository generation ability, providing strong model differentiation while leaving considerable headroom for future progress.

Hantian Ding, Chloe Bi, Jia-Cheng Zhu et al. · 0 citations
Book Open access Oct 2026

From Code to Knowledge: Enhancing Coding Agents via Reverse-Engineered Software Knowledge

Despite advances in code retrieval and static analysis, LLM-based coding agents rarely construct or consult structured representations of a codebase's architectural design, domain model, or behavioral patterns, the very knowledge that human developers depend on when making nontrivial changes. This absence of explicit k...

Yi-Ran Zhang, Yi-Fei Ge, Xiao-Xia Liu · 0 citations
Preprint Oct 2026

Engineering Sustainable Agents: A Systematic Comparison of Agentic LLMs for Developer Workflows

This paper presents a comprehensive empirical study of agentic LLM systems across five software engineering tasks: code generation, technical debt identification, code vulnerability detection, log parsing, and log analysis, and translates these findings into design guidelines for sustainable, task-aware LLM-based devel...

Merve Astekin, Y. N. Tun, Arda Goknil et al. · 0 citations
#natural language process... Preprint Oct 2026

Chronos Enables Code Agents to Reason over Software Evolution

Historical pull requests record the design decisions, compatibility constraints, and implementation patterns behind a codebase's current state. Experience relevant to a new task can span related changes whose descriptions emphasize different concerns. We introduce Chronos, a test-time framework that makes this connecte...

Xin Yin, Yiang Zhang, Zhi-Yuan Peng et al. · 1 citation · ⚡1

Related blog posts

Microsoft Research Blog Nov 12, 2026

Microsoft Research Focus – Edition #98

What's new in research at Microsoft? Host Anson Ho explores the latest, including AI-powered universal translation, a breakthrough in a 31-year-old mathematics challenge, and advances making AI coding agents faster and more efficient.  The post Microsoft Research Focus – Edition #98 appeared first on Microsoft Research.

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.