RELIC: Revealed Principles for Learning Interpretable Composable Skills in Multi-Agent Planning

2026-07-18

AI Summary

This work addresses the challenge of multi-agent planning in privacy-sensitive and heterogeneous settings, where conventional centralized or policy-sharing approaches are infeasible. The authors propose RELIC, a novel framework that decouples coordination from execution for the first time: each agent leverages its private large language model (LLM) to guide procedural skill search, while a trusted coordinator performs black-box evaluations based solely on team-level performance. Successful behaviors are abstracted into interpretable and composable “revelation principles,” which other agents can instantiate and recombine within their local interfaces. This approach enables privacy-preserving skill transfer and compositional learning across agents without exposing internal policies, significantly enhancing planning performance and generalization under heterogeneous interfaces.

Abstract

Multi-agent planning becomes substantially harder when agents must improve specialized decision-making skills while keeping their internal implementations private. This regime arises when agents are developed independently, expose different interfaces and capabilities, and must nevertheless coordinate without sharing executable policies. Prior research has largely assumed centralized optimization, shared policy access, or common skill representations, making it poorly suited to privacy-constrained cooperation. We introduce RELIC, a framework for learning interpretable and composable skills via revealed principles. Each agent refines its own programmatic skill through private LLM-guided search, while a trusted orchestrator evaluates proposed updates solely through team-level performance. Successful behaviors are not broadcast as code; instead, they are abstracted into portable principles that other agents can instantiate within their own interfaces and recombine with local strategies. This separates coordination from implementation sharing, enabling cross-agent transfer under heterogeneous skill signatures. RELIC thus introduces a new paradigm for privacy-preserving skill learning and coordination in multi-agent planning.

Problem

• multi-agent planning
• privacy-preserving coordination
• interpretable skills
• composable skills
• heterogeneous agents

Innovation

• privacy-preserving coordination
• interpretable skills
• composable principles
• heterogeneous multi-agent systems
• LLM-guided skill learning