Skilled AI Agents for Embedded and IoT Systems Development

TL;DR

This paper presents a skills-based AI agent framework and a comprehensive benchmark for developing and evaluating AI systems in embedded and IoT hardware environments, addressing deployment challenges.

Contribution

It introduces a novel skills-based agentic framework and IoT-SkillsBench benchmark for systematic evaluation of AI agents in real embedded hardware settings.

Findings

Human-expert skills achieve near-perfect success rates.

Structured expert knowledge significantly improves AI agent performance.

Benchmark covers diverse platforms, peripherals, and tasks.

Abstract

Large language models (LLMs) and agentic systems have shown promise for automated software development, but applying them to hardware-in-the-loop (HIL) embedded and Internet-of-Things (IoT) systems remains challenging due to the tight coupling between software logic and physical hardware behavior. Code that compiles successfully may still fail when deployed on real devices because of timing constraints, peripheral initialization requirements, or hardware-specific behaviors. To address this challenge, we introduce a skills-based agentic framework for HIL embedded development together with IoT-SkillsBench, a benchmark designed to systematically evaluate AI agents in real embedded programming environments. IoT-SkillsBench spans three representative embedded platforms, 23 peripherals, and 42 tasks across three difficulty levels, where each task is evaluated under three agent configurations (no-skills, LLM-generated skills, and human-expert skills) and validated through real hardware execution. Across 378 hardware validated experiments, we show that concise human-expert skills with structured expert knowledge enable near-perfect success rates across platforms.