Beyond "What to Retrieve": Uncertainty in Retrieval-Augmented Code Generation
2026-07-29T08:51:39Z•54a95db5012a75c0b951f9b5cf85004dd96c07565f1431dc53903cae89f07488
AI-assisted codingDevOpsDockerJavaScriptLLMagent skillsautomated bug findingcode generationcontainer securitydeployment securitydeveloper editsformal verificationnetwork segmentationrepository contextretrieval-augmented generationsecrets managementsecurity requirementssoftware engineeringsupply-chain securitytraceability
What happened
A collection of newly published software-engineering research covering uncertainty-aware retrieval for code generation, JavaScript testability, GenAI-assisted literature reviews, agent skill authoring, developer edits of AI-generated code, specification-driven DevOps, NFR-to-code traceability, autonomous formal verification, and repository-context systems for coding agents. Security-relevant findings include deployment-intent gaps in LLM-generated configurations, difficulty tracing security requirements to implementation, third-party agent-skill trust concerns, and automated discovery of bugs,
Why it matters
A reviewed impact interpretation has not been published for this record.
Evidence and limitations
- Source ID
- arxiv_cs_se
- Record identifier
- 54a95db5012a75c0b951f9b5cf85004dd96c07565f1431dc53903cae89f07488
- Enrichment time
- 2026-07-29T08:51:39Z
- 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.