MOCHA: Multi-Objective Chebyshev Annealing for Agent Skill Optimization

TL;DR

MOCHA introduces a multi-objective optimization method using Chebyshev scalarization and annealing to improve agent skill performance while respecting platform constraints, outperforming existing methods.

Contribution

It presents MOCHA, a novel multi-objective optimization approach that effectively explores Pareto fronts, including non-convex regions, for skill optimization in language model agents.

Findings

MOCHA achieves 7.5% relative improvement in mean correctness over baselines.

MOCHA discovers twice as many Pareto-optimal skill variants.

Existing optimizers fail to improve seed skills in 4 out of 6 tasks.

Abstract

LLM agents organize behavior through skills - structured natural-language specifications governing how an agent reasons, retrieves, and responds. Unlike monolithic prompts, skills are multi-field artifacts subject to hard platform constraints: description fields are truncated for routing, instruction bodies are compacted via progressive disclosure, and co-resident skills compete for limited context windows. These constraints make skill optimization inherently multi-objective: a skill must simultaneously maximize task performance and satisfy platform limits. Yet existing prompt optimizers either ignore these trade-offs or collapse them into a weighted sum, missing Pareto-optimal variants in non-convex objective regions. We introduce MOCHA (Multi-Objective Chebyshev Annealing), which replaces single-objective selection with Chebyshev scalarization - covering the full Pareto front, including non-convex regions - combined with exponential annealing that transitions from exploration to exploitation. In our experiments across six diverse agent skills - where all methods share the same multi-objective mutation operator and baselines receive identical per-objective textual feedback - existing optimizers fail to improve the seed skill on 4 of 6 tasks: 1000 rollouts yield zero progress. MOCHA breaks through on every task, achieving 7.5% relative improvement in mean correctness over the strongest baseline (up to 14.9% on FEVER and 10.4% on TheoremQA) while discovering twice as many more Pareto-optimal skill variants.