From Goals to Aspects, Revisited: An NFR Pattern Language for Agentic AI Systems

Researchers have developed a pattern language methodology to systematically identify and modularize crosscutting concerns in agentic AI systems, addressing issues like security, reliability, and cost management that contribute to high AI project failure rates. The approach uses goal models to discover reusable patterns and implements them through aspect-oriented programming in Rust.

• →High failure rates of AI projects reaching production are partly due to poorly modularized crosscutting concerns like security and fault tolerance.
• →The methodology presents 12 reusable patterns across four categories: security, reliability, observability, and cost management.
• →Four patterns specifically address agent-unique concerns: tool-scope sandboxing, prompt injection detection, token budget management, and action audit trails.
• →The approach extends traditional aspect-oriented programming to the agentic AI domain using Rust implementation.
• →Validation through an open-source autonomous agent framework case study demonstrates systematic identification of crosscutting concerns.