# LLM Agent-Assisted Reverse Engineering with Quantitative Readability Metrics

> Uses a composite Quantitative Readability Score (QRS) to guide agents refining decompiled C. Lexical familiarity, structural simplicity and coding idioms are combined with a binary-structure similarity gate. Experiments compare refinement with and without shell access on 210 simple C binaries.

- Authors: Neil Archibald; Ruben Thijssen
- Venue: arXiv
- Status: Preprint
- Publication / posting date: June 5, 2026
- Date source: arXiv v1 submission date.
- First arXiv submission: June 5, 2026
- arXiv: 2606.06838
- Reviewed arXiv version: v1
- Sources reviewed: 2026-10-19
- Added to the catalog: 2026-10-19
- Original paper: <https://arxiv.org/abs/2606.06838>
- PDF: <https://arxiv.org/pdf/2606.06838v1>
- BibTeX: <https://awesomerea.org/papers/citations/paper-qrs-readability.bib>
- Catalog page: <https://awesomerea.org/papers/paper-qrs-readability/>
- Record JSON: <https://awesomerea.org/papers/paper-qrs-readability/record.json>
- Chinese version: [简体中文](https://awesomerea.org/zh/papers/paper-qrs-readability/index.md)

## Scope and limitations

Structural similarity does not prove semantic equivalence. QRS both guides and evaluates refinement; human readability validation remains future work. The selected synthetic corpus limits generalization. Recovering compilable, behaviorally correct code from difficult binaries is outside the claimed contribution.

## Why it fits this collection

Connects agent feedback design with the readability of recovered code.

## Publication and review notes

Reviewed the v1 abstract and full HTML. No implementation artifact identified.

## Primary sources

1. <https://arxiv.org/abs/2606.06838v1>
2. <https://arxiv.org/html/2606.06838v1>

## About this entry

Reviewed against the paper and its primary sources. Cite the original authors using the linked paper or BibTeX.
Original editorial content: MIT. Linked projects, papers and assets retain their own rights and licenses.