Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

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

Yijun Yu

TL;DR

A pattern language of 12 reusable patterns organized across four NFR categories, each mapping an i* goal model to a concrete aspect implementation using an AOP framework for Rust, offers a principled approach for engineering reliable agentic AI systems through early identification of crosscutting concerns.

Abstract

Agentic AI systems exhibit numerous crosscutting concerns -- security, observability, cost management, fault tolerance -- that are poorly modularized in current implementations, contributing to the high failure rate of AI projects in reaching production. The goals-to-aspects methodology proposed at RE 2004 demonstrated that aspects can be systematically discovered from i* goal models by identifying non-functional soft-goals that crosscut functional goals. This paper revisits and extends that methodology to the agentic AI domain. We present a pattern language of 12 reusable patterns organized across four NFR categories (security, reliability, observability, cost management), each mapping an i* goal model to a concrete aspect implementation using an AOP framework for Rust. Four patterns address agent-specific crosscutting concerns absent from traditional AOP literature: tool-scope sandboxing, prompt injection detection, token budget management, and action audit trails. We extend the V-graph model to capture how agent tasks simultaneously contribute to functional goals and non-functional soft-goals. We validate the pattern language through a case study analyzing an open-source autonomous agent framework, demonstrating how goal-driven aspect discovery systematically identifies and modularizes crosscutting concerns. The pattern language offers a principled approach for engineering reliable agentic AI systems through early identification of crosscutting concerns.

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

TL;DR

A pattern language of 12 reusable patterns organized across four NFR categories, each mapping an i* goal model to a concrete aspect implementation using an AOP framework for Rust, offers a principled approach for engineering reliable agentic AI systems through early identification of crosscutting concerns.

Abstract

Agentic AI systems exhibit numerous crosscutting concerns -- security, observability, cost management, fault tolerance -- that are poorly modularized in current implementations, contributing to the high failure rate of AI projects in reaching production. The goals-to-aspects methodology proposed at RE 2004 demonstrated that aspects can be systematically discovered from i* goal models by identifying non-functional soft-goals that crosscut functional goals. This paper revisits and extends that methodology to the agentic AI domain. We present a pattern language of 12 reusable patterns organized across four NFR categories (security, reliability, observability, cost management), each mapping an i* goal model to a concrete aspect implementation using an AOP framework for Rust. Four patterns address agent-specific crosscutting concerns absent from traditional AOP literature: tool-scope sandboxing, prompt injection detection, token budget management, and action audit trails. We extend the V-graph model to capture how agent tasks simultaneously contribute to functional goals and non-functional soft-goals. We validate the pattern language through a case study analyzing an open-source autonomous agent framework, demonstrating how goal-driven aspect discovery systematically identifies and modularizes crosscutting concerns. The pattern language offers a principled approach for engineering reliable agentic AI systems through early identification of crosscutting concerns.
Paper Structure (31 sections, 7 figures, 6 tables)

This paper contains 31 sections, 7 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Background
  3. From Goals to Aspects (RE 2004)
  4. The $i^*$ Framework
  5. Aspect-Oriented Programming
  6. Agentic AI Systems
  7. Methodology: Goals to Aspects for Agents
  8. Phase 1: Agent Goal Modeling
  9. Phase 2: V-Graph Analysis for Agents
  10. Phase 3: Pattern Instantiation
  11. Pattern Language
  12. Pattern Format
  13. Tool Scope Sandbox (Security, Novel)
  14. Token Budget Manager (Cost, Novel)
  15. Action Audit Trail (Observability, Novel)
  16. ...and 16 more sections

Figures (7)

  • Figure 1: Overlapping V-graphs for agents: the "Call LLM" task negatively impacts four NFR softgoals (red dashed arrows), each revealing a crosscutting concern requiring a mitigating aspect.
  • Figure 2: Three-phase methodology for discovering aspects from agent goal models.
  • Figure 3: Pattern relationships: solid arrows show prerequisites, dashed arrows show composition relationships. Patterns marked $\star$ are novel.
  • Figure 4: ZeroClaw codebase growth (LOC, bars) and community adoption (GitHub stars, line) across four analysis snapshots. The framework grew 3.4$\times$ in seven days while accumulating 18,000 GitHub stars, indicating that the crosscutting concerns identified here affect a rapidly growing, widely-adopted system.
  • Figure 5: Strategic Dependency model for ZeroClaw with five actors and their goal dependencies.
  • ...and 2 more figures

Theorems & Definitions (2)

  • Definition 1: V-graph for Agents
  • Definition 2: Crosscutting Density