LedgerAgent: Structured State for Policy-Adherent Tool-Calling Agents

2026-06-18

AI Summary

This work addresses the limitations of existing tool-calling agents in customer service scenarios, which often make erroneous decisions or violate domain-specific policies due to outdated or missing information stemming from the absence of explicit task state management. To overcome this, we propose LedgerAgent, the first agent framework to incorporate an explicit state ledger mechanism that structurally records task states during reasoning and injects relevant state information into prompt generation. Furthermore, before executing environment-altering tools, LedgerAgent validates policy constraints dependent on current state, thereby decoupling task state management from policy enforcement. Experimental results across four customer service domains and multiple large language models demonstrate that our approach significantly improves task success rates, with particularly notable gains in multi-turn consistency metrics.

Abstract

Policy-adherent tool-calling agents in customer-service domains must maintain task states across turns while calling tools and obeying domain policies. Task states consist of relevant facts, identifiers, constraints, and conditions observed through user interaction and tool calls. In standard agents, task states are not represented separately. Observations, tool returns, and policy instructions are placed in the prompt, leaving agents to reconstruct the relevant states from the prompt each time they decide what to do next. This design makes state management implicit, creating two common failure modes. An agent may retrieve the right facts but later ground its decision in stale, missing, or incorrect information; and a syntactically valid tool call may still violate a domain policy that depends on the current task state. We introduce \textsc{LedgerAgent}, an inference-time method for tool-calling agents that maintains observed task states in a separate ledger and renders the states into the prompt. The ledger is also used to check state-dependent policy constraints before environment-changing tool calls are executed, blocking policy violations. Across four customer-service domains and a mixed panel of open- and closed-weight models, \textsc{LedgerAgent} improves average pass\textasciicircum{}k over a standard prompt-based tool-calling approach, with the largest gains under stricter multi-trial consistency metrics.

Problem

• tool-calling agents
• task state management
• policy adherence
• customer-service domains
• state-dependent constraints

Innovation

• structured state
• policy adherence
• tool-calling agents
• state ledger
• inference-time method