AgentFactory: A Self-Evolving Framework Through Executable Subagent Accumulation and Reuse

2026-03-18

AI Summary

This work proposes a novel self-evolution framework for large language model (LLM) agents that addresses the limitations of existing approaches, which rely on textual experience logs that are difficult to reliably reuse and efficiently execute in complex tasks. Instead of storing experiences as natural language, the proposed method preserves successful solutions as executable Python sub-agents. These sub-agents are continuously refined through execution feedback, enabling automatic capability accumulation and cross-system reuse. By leveraging executable code as the medium of experience—combined with modular design and standardized interfaces—the framework constructs an evolving library of sub-agents that improves over time. This paradigm significantly reduces the computational and cognitive overhead for handling similar tasks. The implementation has been open-sourced to facilitate further research and adoption.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Multiagent Systems: Adversarial Agents
• Natural Language Processing: (Large) Language Models

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Building LLM-based agents has become increasingly important. Recent works on LLM-based agent self-evolution primarily record successful experiences as textual prompts or reflections, which cannot reliably guarantee efficient task re-execution in complex scenarios. We propose AgentFactory, a new self-evolution paradigm that preserves successful task solutions as executable subagent code rather than textual experience. Crucially, these subagents are continuously refined based on execution feedback, becoming increasingly robust and efficient as more tasks are encountered. Saved subagents are pure Python code with standardized documentation, enabling portability across any Python-capable system. We demonstrate that AgentFactory enables continuous capability accumulation: its library of executable subagents grows and improves over time, progressively reducing the effort required for similar tasks without manual intervention. Our implementation is open-sourced at https://github.com/zzatpku/AgentFactory, and our demonstration video is available at https://youtu.be/iKSsuAXJHW0.

Problem

• LLM-based agents
• self-evolution
• executable subagents
• task re-execution
• complex scenarios

Innovation

• executable subagents
• self-evolution
• LLM-based agents
• code reuse
• capability accumulation