ExComm: Exploration-Stage Communication for Error-Resilient Agentic Test-Time Scaling

2026-05-21

AI Summary

This work addresses the challenge of belief contamination and error propagation in long-horizon multi-agent systems during testing, which often arises from intermediate mistakes. To mitigate this issue, the authors propose ExComm, a novel communication protocol that introduces, for the first time, a conflict-aware mechanism during the exploration phase. ExComm periodically audits agents’ belief states to detect cross-agent factual inconsistencies and integrates a tool-verification loop for error correction. Corrected information is then softly incorporated into agents’ beliefs, while a trajectory diversification module preserves policy diversity. Experimental results on the AIME 2024, AIME 2025, and GAIA benchmarks demonstrate that, when paired with Gemini-2.5-Flash-Lite and Qwen3.5-4B models, ExComm outperforms the strongest baseline by average margins of 5.7% and 5.0%, respectively, significantly enhancing error recovery, reasoning diversity, and cost efficiency.

Abstract

A common failure mode in long-horizon agentic test-time scaling is error propagation, where factual errors or invalid deductions introduced at intermediate steps persist in the agent's belief state and contaminate later reasoning. Existing test-time scaling methods provide limited control over this process, as they often rely on agents to detect their own mistakes, select among flawed trajectories, or refine solutions only after errors have already shaped the reasoning path. We propose ExComm, a communication protocol for exploration-stage agentic test-time scaling. ExComm is motivated by the empirical observation that the majority of intermediate errors in parallel agentic reasoning produce detectable cross-agent factual conflicts. Leveraging the iterative structure of agentic workflows, ExComm periodically audits agent belief states to detect such conflicts, resolves them through a dedicated tool-based verification loop, and returns concise, targeted feedback to the involved agents. Corrections are incorporated through soft belief updates, which append verified feedback rather than overwriting existing beliefs. Furthermore, to prevent collapsing trajectory diversity due to communication, ExComm further introduces a trajectory diversification module that redirects redundant trajectories toward orthogonal strategies. Experiments on AIME 2024, AIME 2025, and GAIA with Gemini-2.5-Flash-Lite and Qwen3.5-4B show that ExComm consistently outperforms strong test-time scaling baselines, achieving average performance gains of 5.7% and 5.0% over the best-performing baselines, respectively. Further analyses demonstrate improved error recovery, favorable scaling behavior, stronger diversity than adapted communication baselines, and the best performance-cost trade-off among the evaluated methods.

Problem

• error propagation
• agentic reasoning
• test-time scaling
• belief state contamination
• long-horizon reasoning

Innovation

• error-resilient agentic reasoning
• exploration-stage communication
• belief conflict detection
• soft belief update
• trajectory diversification