Saved in:
Bibliographic Details
Main Authors: Chen, Qijia, Bellucci, Andrea, Sun, Zhida, Jacucci, Giulio
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.14872
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918450600869888
author Chen, Qijia
Bellucci, Andrea
Sun, Zhida
Jacucci, Giulio
author_facet Chen, Qijia
Bellucci, Andrea
Sun, Zhida
Jacucci, Giulio
contents LLM-based mobile GUI agents treat every task invocation as an independent reasoning episode, requiring a full LLM inference call at each action step. This per-step dependence makes them stateless: a task completed successfully yesterday is re-derived from scratch today, with no improvement in reliability or speed. We present SkillDroid, a three-layer skill agent that compiles successful LLM-guided GUI trajectories into parameterized skill templates (sequences of UI actions with weighted element locators and typed parameter slots) and replays them on future invocations without any LLM calls. A matching cascade (regex patterns, embedding similarity, and app filtering) routes incoming instructions to stored skills, while a failure-learning layer triggers recompilation when skill reliability degrades. Over a 150-round longitudinal evaluation with systematic instruction variation and controlled perturbations, SkillDroid achieves an 85.3% success rate (23 percentage points above a stateless LLM baseline) while using 49% fewer LLM calls. The skill replay mechanism achieves a perfect 1000% success rate across 79 replay rounds at 2.4 times the speed of full LLM execution. Most critically, the system improves with use: its success rate converges upward from 87% to 91%, while the baseline degrades from 80% to 44%.
format Preprint
id arxiv_https___arxiv_org_abs_2604_14872
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle SkillDroid: Compile Once, Reuse Forever
Chen, Qijia
Bellucci, Andrea
Sun, Zhida
Jacucci, Giulio
Human-Computer Interaction
LLM-based mobile GUI agents treat every task invocation as an independent reasoning episode, requiring a full LLM inference call at each action step. This per-step dependence makes them stateless: a task completed successfully yesterday is re-derived from scratch today, with no improvement in reliability or speed. We present SkillDroid, a three-layer skill agent that compiles successful LLM-guided GUI trajectories into parameterized skill templates (sequences of UI actions with weighted element locators and typed parameter slots) and replays them on future invocations without any LLM calls. A matching cascade (regex patterns, embedding similarity, and app filtering) routes incoming instructions to stored skills, while a failure-learning layer triggers recompilation when skill reliability degrades. Over a 150-round longitudinal evaluation with systematic instruction variation and controlled perturbations, SkillDroid achieves an 85.3% success rate (23 percentage points above a stateless LLM baseline) while using 49% fewer LLM calls. The skill replay mechanism achieves a perfect 1000% success rate across 79 replay rounds at 2.4 times the speed of full LLM execution. Most critically, the system improves with use: its success rate converges upward from 87% to 91%, while the baseline degrades from 80% to 44%.
title SkillDroid: Compile Once, Reuse Forever
topic Human-Computer Interaction
url https://arxiv.org/abs/2604.14872