Constraint Decay: The Fragility of LLM Agents in Backend Code Generation

2026-05-07

AI Summary

This study addresses the challenge large language model (LLM) agents face in simultaneously satisfying functional correctness and strict structural constraints—such as architectural patterns, database schemas, and ORM usage—when generating backend code. The authors construct a standardized multi-file benchmark spanning Flask, FastAPI, and Django, evaluating LLM performance on 80 zero-shot generation and 20 functionality extension tasks through end-to-end behavioral testing and static structural validation. They identify and name a novel phenomenon, “constraint decay”: as structural constraints intensify, assertion pass rates for advanced configurations drop by an average of 30 percentage points, with weaker setups approaching zero. Framework complexity significantly impacts outcomes, and data-layer defects—including erroneous queries and ORM violations—emerge as primary failure sources, revealing critical limitations of current code-generation agents in production-like environments.

Abstract

Large Language Model (LLM) agents demonstrate strong performance in autonomous code generation under loose specifications. However, production-grade software requires strict adherence to structural constraints, such as architectural patterns, databases, and object-relational mappings. Existing benchmarks often overlook these non-functional requirements, rewarding functionally correct but structurally arbitrary solutions. We present a systematic study evaluating how well agents handle structural constraints in multi-file backend generation. By fixing a unified API contract across 80 greenfield generation tasks and 20 feature-implementation tasks spanning eight web frameworks, we isolate the effect of structural complexity using a dual evaluation with end-to-end behavioral tests and static verifiers. Our findings reveal a phenomenon of constraint decay: as structural requirements accumulate, agent performance exhibits a substantial decline. Capable configurations lose 30 points on average in assertion pass rates from baseline to fully specified tasks, while some weaker configurations approach zero. Framework sensitivity analysis exposes significant performance disparities: agents succeed in minimal, explicit frameworks (e.g., Flask) but perform substantially worse on average in convention-heavy environments (e.g., FastAPI, Django). Finally, error analysis identifies data-layer defects (e.g., incorrect query composition and ORM runtime violations) as the leading root causes. This work highlights that jointly satisfying functional and structural requirements remains a key open challenge for coding agents.

Problem

• structural constraints
• backend code generation
• LLM agents
• constraint decay
• non-functional requirements

Innovation

• constraint decay
• structural constraints
• LLM agents
• backend code generation
• framework sensitivity