Multi-agent Collaboration with State Management

2026-05-19

AI Summary

This work addresses the challenge of silent conflicts and inconsistent code views that arise when multiple agents concurrently edit a shared codebase, often leading to integration failures. To resolve this, the authors propose STORM (State-Oriented Management), a novel mediation architecture that abandons traditional workspace isolation in favor of state-oriented coordination. STORM detects and resolves conflicts in real time during the write phase, ensuring all agents operate on a consistent view of the codebase. The approach is agnostic to both multi-agent system design and large language model choice, enabling seamless integration into existing software development workflows. Experimental results demonstrate that STORM outperforms baseline methods by 18.7 and 1.4 points on Commit0-Lite and PaperBench benchmarks, achieving peak scores of 87.6 and 78.2, respectively, while maintaining comparable or superior cost efficiency.

Abstract

Recent advances in multi-agent systems have shown great potential for solving complex tasks. However, when multiple agents edit a shared codebase concurrently, their changes can silently conflict and inconsistent views lead to integration failures. Existing multi-agent systems address this through workspace isolation (e.g., one git worktree per agent), but this defers conflict resolution to a post-hoc merge step where recovery is expensive. In this paper, we propose STORM, i.e., STate-ORiented Management for multi-agent collaboration. Specifically, STORM manages agent states by mediating their interactions with the shared workspace, ensuring that each agent operates on a consistent view of the codebase and that conflicting edits are detected and resolved at write time. We evaluate STORM on Commit0 and PaperBench across multiple LLMs. STORM outperforms the git-worktree-based multi-agent baseline by +18.7 on Commit0-Lite and +1.4 on PaperBench, while achieving comparable or better cost efficiency. Combined with single-agent runs, STORM reaches highest scores of 87.6 and 78.2 on the two benchmarks respectively, suggesting that explicit state management is a more effective foundation for multi-agent collaboration than workspace isolation. STORM can also be plugged into any multi-agent system seamlessly.

Problem

• multi-agent collaboration
• state management
• codebase conflicts
• shared workspace
• integration failures

Innovation

• multi-agent collaboration
• state management
• conflict resolution
• shared workspace
• consistent view