StateFuse: Deterministic Conflict-Preserving Memory for Multi-Agent Systems

2026-07-07

AI Summary

This work addresses the limitations of existing memory layers in multi-agent systems, which often obscure observation conflicts through overwriting, thereby hindering auditability and correction. The authors propose a conflict-aware replicated memory contract grounded in standard OpSet/CRDT merging semantics. Without introducing new merge algebras, their approach leverages immutable history, explicit conflict objects, semantic handles via claim_id/claim_ref, deterministic predicate contracts, and resolution-at-projection mechanisms to ensure conflict visibility, safe abstention, and auditable corrections. Experimental evaluation on MemoryAgentBench demonstrates that the method maintains accuracy comparable to existing approaches while significantly enhancing contradiction retention. Notably, when precise identifiers are unavailable, the semantic handles prove essential for reliable correction.

Abstract

Agent systems accumulate conflicting observations across branches, retries, and replicas, yet many practical memory layers still collapse disagreement behind overwrite rules that are difficult to inspect or correct. We present StateFuse, a conflict-aware replicated memory contract built on standard OpSet/CRDT merge. StateFuse does not introduce a new join algebra; it defines an agent-facing semantics layer with immutable history, explicit conflict objects, exact and semantic correction handles (claim_id / claim_ref), deterministic predicate contracts, and projection-time resolution that cannot rewrite replicated state. We evaluate StateFuse against flat multi-value, raw-log, provenance-style, and collapsed baselines under matched resolver and verification policies. On a 282-question official conflict-bearing MemoryAgentBench slice, the compared methods tie on answer accuracy, but conflict-preserving surfaces keep contradictions visible while collapsed surfaces do not. In a controlled agent loop with uniform verification, preserving ambiguity enables safer abstention and correction than early collapse. A correction-handle ablation further shows that semantic handles matter when exact prior identifiers are unavailable. The resulting claim is narrow: StateFuse is best supported as a safer public memory contract for contradiction surfacing, abstention, and auditable correction, not as a universal accuracy gain.

Problem

• multi-agent systems
• conflicting observations
• replicated memory
• conflict preservation
• memory consistency

Innovation

• conflict-preserving memory
• multi-agent systems
• CRDT
• deterministic semantics
• auditable correction