MOSCOPT: Mixture-of-Skills Collective Optimization for LLM Agents

arXiv:2609.14399v1 Announce Type: new Abstract: Natural language prompts and skills serve as the strategic backbone of LLM-based agents. Recent advances in prompt and skill optimization have achieved notable gains, yet all existing methods optimize a single text template---missing the synergy among multiple complementary strategies. We propose MOSCOPT, a text-native, parameter-free algorithm that jointly optimizes a pool of N skills and a gating skill G that dynamically selects K skills per step. To effectively optimize the skills, we build the EditAdam with internally maintained dual states. Through the three-phase interleaved updates with EditAdam, the system monotonically improves without gradient or parameter tuning. Extensive experiments and detailed ablations across 5 benchmarks and 3 target LLMs demonstrate that MOSCOPT consistently outperforms all baselines, and confirm that both the mixture-of-skills architecture with selective activation and the collective evolution with three-phase interleaving are essential to its superior performance. Code is released https://github.com/zhangzhenyu13/SummerClaw/tree/master/summerclaw/agent_trainer/algorithms/moscopt.

Related

• Bidirectional Context Self-Distillation for Reinforcement Learning of Skill-Based LLM Agents
• COBRA-Skills: Contextual Bandit-Guided Evolution for Agent Skill Optimization
• OptSkills: Learning Generalizable Optimization Skills from Problem Archetypes via Cluster-Based Distillation
• SkillAdaptor: Self-Adapting Skills for LLM Agents from Trajectories

Source: arXiv cs.AI | 2026-09-15