Evolving Programmatic Skill Networks

2026-01-07

arXiv.org

AI Summary

This work addresses the challenge of continual learning, optimization, and reuse of executable skills for agents operating in open-ended embodied environments. The authors propose the Procedural Skill Network (PSN), which constructs a composable and evolvable skill library using executable symbolic programs. PSN integrates three large language model–based mechanisms: structured failure localization (REFLECT), maturity-aware progressive update gating, and structure reconstruction under rollback validation. Together, these mechanisms effectively balance learning stability and plasticity. Experimental results demonstrate that PSN significantly enhances skill reusability, adaptation speed, and cross-task generalization performance in benchmark environments such as MineDojo and Crafter.

Technology Category

• Multiagent Systems: Multiagent Learning
• Machine Learning: Neuro-Symbolic Learning
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: Social networks and social learning

Abstract

We study continual skill acquisition in open-ended embodied environments where an agent must construct, refine, and reuse an expanding library of executable skills. We introduce the Programmatic Skill Network (PSN), a framework in which skills are executable symbolic programs forming a compositional network that evolves through experience. PSN defines three core mechanisms instantiated via large language models: (1)REFLECT for structured fault localization over skill compositions, (2) progressive optimization with maturity-aware update gating that stabilizes reliable skills while maintaining plasticity for uncertain ones, and (3) canonical structural refactoring under rollback validation that maintains network compactness. We further show that PSN's learning dynamics exhibit structural parallels to neural network training. Experiments on MineDojo and Crafter demonstrate robust skill reuse, rapid adaptation, and strong generalization across open-ended task distributions.\footnote{We plan to open-source the code.

Problem

• continual skill acquisition
• open-ended embodied environments
• executable skills
• skill reuse
• compositional skill networks

Innovation

• Programmatic Skill Network
• continual skill acquisition
• structured fault localization
• maturity-aware update gating
• structural refactoring