The Semi-Executable Stack: Agentic Software Engineering and the Expanding Scope of SE

Dimensions overall score 1.0

• 31(2), 2022. doi: 10.1145/3487043. [28] Microsoft WorkLab. 2025: The year the frontier firm is born.https://www.microsoft.co m/en-us/worklab/work-trend-index/2025-the-year-the-frontier-firm-is-born , 2025. Published April 29, 2025. [29] Model Evaluation and Threat Researchclaim 82%
• provide useful maps of the current technical landscape [8, 17, 26]. Their center of gravity lies in rings 1–3: code generation, bug fixing, testing, repair, prompt engineering, agent design, and benchmark-driven capability progress. Benchmarks such as SWE-bench extend that frontier toward longer-horizonclaim 82%
• neighbors for ring 4. Their main contribution is to treat deployment, monitoring, reproducibility, and operational control as first-class concerns rather than afterthoughts [22]. The stack agrees with that move, but pushes the boundary further outward. MLOps usually centers the operation of models and pipelinesclaim 82%
• parsing problem. The contribution’s limits sit at a different ring from its technical strength. The pipeline improves a coding-adjacent activity but leaves ring-4 and ring-5 questions open: who trusts its outputs, how those outputs enter release decisions, how exceptions are handled, and how surrounding roles changelimitation 82%