Evo2Team: When Do Evolved Skills Transfer? From Selection to Deployment

2026-09-28

AI Summary

This study addresses the uncertainty and evaluation challenges associated with skill transfer in multi-agent systems by proposing Evo2Team, a framework that optimizes target team deployment through the selection, adaptation, and validation of source skills. It innovatively introduces a multi-objective function encompassing joint quality, cost, and model hierarchy, while emphasizing an execution-based transfer evaluation mechanism grounded in actual agent behaviors. The proposed method leverages GPT/Qwen model cascades, Count-Frequency and AgentsNet environments, and evolutionary algorithms for optimization. Experimental results demonstrate that Evo2Team effectively reduces exploration costs across most scenarios, significantly enhances task performance, and lowers deployment overhead.

Abstract

A skill bank that helps one multi-agent system may leave another's behavior unchanged. A transferred rule helps only when target agents act on it successfully. We study this path for routing and communication skills in Count-Frequency and AgentsNet, using teams of 4--32 agents and GPT and Qwen model ladders. Source evolution meets a joint quality, cost, model-tier, and confirmation goal in 14 of 16 settings. We then evaluate Evo2Team, which selects, adapts, and confirms source skills for the target team, alongside six frozen selectors across 28 transfer directions. Evo2Team's target-side exploration cost is below that of evolving a new target bank in every direction, even when reused reference evaluations are charged once. Twenty of 28 held-out outcomes meet the positive-transfer criterion, including three saved diagnostic tests. Selection alone does not explain these outcomes: KNN and CORAL choose different banks in two AgentsNet directions but produce identical recorded executions. When Evo2Team changes execution, gains can reach many tasks, as in a Count-Frequency direction that improves 28 of 32 tasks over KNN. Seven positive AgentsNet outcomes save 6.1--14.6\% in deployment cost while using transferred skills on only three to six of fifteen tasks. In five earlier accepted directions, all 22 task records using transferred skills pass three fixed-graph confirmations, but four fail in recorded executions on new graphs. Graphs and model responses change together in this comparison. These results show that skill transfer must be assessed through the actions agents take, the tasks those actions reach, and the quality and cost of the final deployment.

Problem

• skill transfer
• multi-agent systems
• positive transfer
• deployment cost
• evolved skills

Innovation

• Skill Transfer
• Multi-Agent Systems
• Evolutionary Selection
• Deployment Cost
• Cross-team Adaptation