ToolAnchor: Anchoring Counterfactual Context to Boost Agentic Tool-use Capability

2026-07-14

AI Summary

This work addresses the challenge that large language model (LLM) agents often struggle to effectively leverage newly introduced tools due to entrenched behavioral inertia from prior experience. To overcome this limitation, the authors propose ToolAnchor, a novel framework that introduces counterfactual anchored contexts as an intervention mechanism to disrupt habitual decision-making at critical points, enabling dynamic tool integration without requiring full retraining. ToolAnchor employs a teacher–student architecture to generate and validate counterfactual contexts, complemented by agent post-training and trajectory rollback mechanisms to refine decision policies. Experimental results demonstrate that ToolAnchor significantly improves task success rates across multiple benchmarks—including GAIA, BrowseComp, and VDR-Bench—when agents operate with expanded tool sets.

Abstract

Tool-augmented large language model agents excel at long-horizon tasks, yet they are typically post-trained on fixed toolsets. When tasks demand new tools, these agents struggle to incorporate them effectively, and retraining from scratch is often impractical. We identify the core obstacle in such toolset expansion problem as behavioral inertia: the tendency of agents to fall back on familiar tools and established reasoning patterns despite having access to new ones. We demonstrate that injecting counterfactual anchor contexts at critical decision points can break this inertia, recovering failed trajectories by eliciting suppressed agent capabilities. To scale this insight, we propose ToolAnchor, a framework that uses teacher models to hypothesize these counterfactual contexts, verifies them via student rollouts, and internalizes the successful interventions through agentic post-training. Extensive evaluations across general AI assistant (GAIA), textual search (BrowseComp), and visual search (VDR-Bench) tasks demonstrate that ToolAnchor consistently exhibits competitive performance under expanded toolsets. Our work bridges the gap between static post-training and dynamic adaptation, charting a new path for scalable agentic reinforcement learning.

Problem

• tool-augmented agents
• behavioral inertia
• toolset expansion
• counterfactual context
• agentic tool-use

Innovation

• ToolAnchor
• counterfactual context
• behavioral inertia
• agentic tool-use
• post-training adaptation