Context Before Code: An Experience Report on Vibe Coding in Practice

2026-03-10

AI Summary

This work addresses a critical gap in existing code generation tools, which often neglect essential architectural requirements in production environments—such as multi-tenancy isolation, access control, and infrastructure constraints. The authors propose a “context-over-code” paradigm that explicitly embeds architectural constraints (e.g., role-based access control, project-level isolation, retrieval-augmented generation for provenance, structured memory, and asynchronous processing) into prompts to guide large language models in constructing maintainable, multi-project agent platforms and academic retrieval-augmented systems. By shifting engineering focus from boilerplate implementation to constraint specification and auditability, this approach delineates architectural “non-delegable zones” unsuitable for model delegation. Empirical results demonstrate that the strategy significantly accelerates scaffolding and integration while ensuring multi-tenant security, isolation, and runtime reliability, thereby validating the pivotal role of explicit architectural constraints in enhancing the production readiness of generated code.

Technology Category

• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications

Abstract

Code-generating tools are increasingly used in software development, yet experience reports on conversational "vibe coding" under production constraints remain limited. This paper presents an experience report from a small full-stack team that applied contextual prompting and explicit architectural constraints to build (i) a multi-project agent learning platform designed for sustained, production-oriented use and (ii) an academic retrieval-augmented generation system. The agent platform supports multiple isolated projects, each with structured memory and background processing, thereby enforcing project-level isolation. The RAG system provides citation-grounded answers, role-based access control, and evaluation tracking. Across both systems, vibe coding accelerated scaffolding and integration. However, the generated code often under-specified isolation rules and infrastructure constraints when these were not explicitly defined. Consequently, aspects such as multi-tenancy, access control, memory policies, and asynchronous processing required deliberate architectural design and verification. We observe a shift in engineering effort from boilerplate implementation toward constraint specification and enforcement auditing. We also identify recurring architectural "non-delegation zones" where conversational code generation remains insufficient for production reliability.

Problem

• vibe coding
• production constraints
• architectural constraints
• code generation
• multi-tenancy

Innovation

• vibe coding
• contextual prompting
• architectural constraints
• production reliability
• non-delegation zones