AI Paper Atlas

Research map / Multi-agent systems and games / LLM agents

From Local Skills to Long-Horizon Tasks: Progressive Skill Exploration for LLM Web Agents

Xueyang Feng, Xiaohe Bo, Xu Chen, Quanyu Dai, Feng Wen

NeurIPS 2026 · Multi-agent systems and games › LLM agents · approach: LLM multi-agent decision making

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes a progressive skill exploration approach for training LLM web agents, moving from local skills to long-horizon tasks to address sparse rewards and rare successful trajectories.

Problem

Training LLM web agents with naive exploration is hard because successful trajectories are rare and sparse terminal rewards give little useful feedback from failed trajectories.

Approach

Draws on skill acquisition theory to train agents progressively, first exploring local skills and then composing them into long-horizon web tasks, so that feedback is available beyond terminal rewards.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
train with SCOPE progressive exploration instead of WebRLLlama3.1-8B · WebArena-Litetotal success rate: 45.8 -> 54.3Improved
train with SCOPE instead of RFTQwen2.5-3B · WebArena-Litetotal success rate: 36.5 -> 41.6Improved
train with SCOPE instead of RFTLlama3.1-8B · WebVoyagertotal success rate: 37.3 -> 42.8Improved
train with SCOPE instead of GRPOLlama3.1-8B · WebShopsuccess rate: 76.2 -> 85.2Improved
train with SCOPE instead of GRPOQwen2.5-3B · WebShopsuccess rate: 80.8 -> 84.6Improved
use teacher intervention (OPTI) vs self intervention (OPSI) (ablation)SCOPE · WebArena-Lite, Qwen2.5-3Btotal success rate: OPTI 41.6 vs OPSI 40.2Mixed
use self intervention (OPSI) vs teacher intervention (OPTI) (ablation)SCOPE · WebArena-Lite, Llama3.1-8Btotal success rate: OPSI 54.3 vs OPTI 53.6Mixed
add more outer training rounds (1 to 3) (ablation)SCOPE · WebVoyagersuccess rate: total gain of 4.25%Improved
add more inner exploration iterations (ablation)SCOPE · WebVoyagersuccess rate: 33.40% -> 43.41%Improved
use self-managed memory model instead of external DeepSeek-V3.2 (ablation)SCOPE · WebArena/WebVoyagerfinal improvementNo clear effect
remove auxiliary information c (ablation)SCOPE (OPTI) · WebArena-Lite, Llama3.1-8Btotal success rate: 53.6 -> 52.7No clear effect
remove skill discovery, orchestration or expansion (ablation)SCOPE · component ablationsuccess rateImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy