DevIntent: How Much Does LLM-Generated Code Violate Developer Intent?
2026-08-11T08:51:40Z•e49cdf08e8c66ee367b90e9fd1e7ba188a3887f26def244e992f0d6f48daff6a
AI code reliabilityLLM-generated codeagent safetyarXivautomated testingautomotive ADAScode correctnesscode reusecyber-physical systemsdeveloper intentdirected test generationfuzzingintent violationintrusion detectionlarge language modelsmodel-based systems engineeringmulti-agent systemsruntime feedbackself-evolving agentssoftware engineeringsoftware testingtest coverageverification
What happened
This document is an arXiv software-engineering research feed covering LLM-generated code intent violations, runtime-guided test generation, verification-driven and multi-agent engineering workflows, model selection, cyber-physical-system behavior modeling, code-correctness representations, self-evolving coding agents, Stack Overflow code reuse, and testing practices for LLM-based agents. The findings highlight reliability, verification, testing, traceability, and operational-safety challenges in AI-assisted software development. No specific exploitable vulnerability or affected product is**;**
Why it matters
A reviewed impact interpretation has not been published for this record.
Evidence and limitations
- Source ID
- arxiv_cs_se
- Record identifier
- e49cdf08e8c66ee367b90e9fd1e7ba188a3887f26def244e992f0d6f48daff6a
- Enrichment time
- 2026-08-11T08:51:40Z
- AI-assisted enrichment
- Yes
This record may overlap with other records. Its enrichment can be incomplete or wrong, and machine assistance was used. Validate consequential decisions against the linked source and your own environment.