Refusal Evaluation in Coding LLMs and Code Agents: A Systematic Review of Thirteen Malicious-Code Prompt Corpora (2023-2025)

TL;DR

This systematic review analyzes thirteen publicly released prompt corpora for evaluating large language models' refusal to engage in malicious coding tasks, highlighting methodological gaps and proposing directions for future research.

Contribution

It uniquely treats prompt datasets as the primary unit of analysis and provides a comprehensive synthesis of their construction, taxonomy, and validation methods.

Findings

Identified lack of human-annotator baselines for calibration.

Highlighted absence of cross-corpus comparability in refusal rates.

Noted fragmentation in malware-category taxonomies.

Abstract

The evaluation of large language model refusal on malicious-coding tasks now spans at least thirteen publicly released prompt corpora (AdvBench, the CyberSecEval family, RMCBench, RedCode, MCGMark, JailbreakBench, CySecBench, MalwareBench, CIRCLE, MOCHA, ASTRA, Scam2Prompt / Innoc2Scam-bench, and JAWS-Bench), each constructed under a different protocol, released under different licensing terms, and validated (or not) against different inter-rater reliability standards. Existing surveys treat code security, jailbreak taxonomy, or vulnerability detection as the central object and mention these corpora only in passing. This paper reverses that framing: it treats the prompt datasets themselves as the unit of analysis. Following a PRISMA-style protocol, we specify a search strategy, screen the recent literature on coding-LLM refusal evaluation, apply a uniform extraction template to each in-scope corpus, and synthesize the resulting catalogue along construction methodology, prompt-construction taxonomy (modality, turn structure, elicitation style), reproducibility and licensing, and malware-category coverage. The synthesis surfaces three recurring methodological gaps: the absence of human-annotator baselines against which LLM-judge labels can be calibrated, the absence of cross-corpus comparability with refusal-rate statistics measuring non-equivalent constructs, and the fragmentation of malware-category taxonomies, with no canonical schema spanning the thirteen in-scope corpora. The review concludes with proposed methodological directions for next-generation corpora, including pre-registration of inclusion criteria, vendor-diverse multi-judge validation, Fleiss' kappa with bootstrap CI as the reliability baseline, and a candidate canonical taxonomy.