arXiv cs.AISaturday · May 23, 2026FREE

Declarative Data Services: Structured Agentic Discovery for Composing Data Systems

arXiv paper 2605.20690 proposes Declarative Data Services (DDS), an architecture for structured agentic discovery of data-system compositions from declarative user intent. The framework addresses the challenge of LLM-driven search for multi-system data backends, where unbounded agentic discovery—a coding agent iterating on failure-log feedback—fails to converge consistently. DDS introduces four typed contracts at successive layers: intent, operator DAG, per-system skills, and runtime attribution. These decompose the global search into bounded sub-searches, with sub-agents searching each typed space. Knowledge flows forward as inline skill citations, and errors route backward as typed signals. As a proof of life on a trading-backend workload, DDS converges where unbounded discovery does not, even when iteration and explicit composition knowledge are added. The paper is available on arXiv as of May 22, 2026.