Development of a CASE Platform with Generative Artificial Intelligence and an MCP Agent Layer

Development of a CASE Platform with Generative Artificial Intelligence and an MCP Agent Layer

Polygalov B.A. (NRU HSE, Moscow, Russia)
Lanin V.V. (NRU HSE, Moscow, Russia)

Abstract

This paper presents the architecture, implementation and preliminary service-level evaluation of a prototype CASE platform that uses generative artificial intelligence to support requirements structuring, code generation, test-artifact generation and deployment checks. The platform combines an adapter-based LLM integration layer, versioned artifact storage, explicit human approval checkpoints and a bidirectional Model Context Protocol (MCP) layer for tool-oriented integration. The work is positioned against recent research on LLM-based software engineering, test generation, software agents and tool-augmented language models. The evaluation is intentionally limited and does not claim productivity gains or industrial readiness. It uses five web-application scenarios executed in deterministic mode, six success criteria, backend regression tests, a workflow-property comparison with two baselines and an additional exploratory check with a real commercial LLM provider on two project prompts. The results provide preliminary evidence that the prototype can execute a traceable requirements-to-artifacts workflow, generate structurally valid deployment bundles and preserve run metadata. The evaluation does not measure semantic correctness of generated business logic and should be extended with repeated real-provider experiments, larger benchmarks and controlled user studies.

Keywords

case system; generative artificial intelligence; software engineering; LLM; requirements engineering; code generation; test generation; mcp; software architecture; human-in-the-loop; evaluation.

Edition

Proceedings of the Institute for System Programming, vol. 38, issue 4, part 2, 2026, pp. 279-292

ISSN 2220-6426 (Online), ISSN 2079-8156 (Print).

DOI: 10.15514/ISPRAS-2026-38(4)-32

For citation

Polygalov B.A., Lanin V.V. Development of a CASE Platform with Generative Artificial Intelligence and an MCP Agent Layer. Proceedings of the Institute for System Programming, vol. 38, issue 4, part 2, 2026, pp. 279-292 DOI: 10.15514/ISPRAS-2026-38(4)-32.

Full text of the paper in pdf Back to the contents of the volume