Structured Authority Injection: Persistent Human Control for LLM Agents via Typed MCP Overrides

Computer Science > Artificial Intelligence arXiv:2610.22373 (cs) [Submitted on 27 Jul 2026] Title: Structured Authority Injection: Persistent Human Control for LLM Agents via Typed MCP Overrides Authors: Ryan Tran, YueMing Zhang, Lingfeng Tao, Nan Yang, Shengjie Xu, Yu Yin, Shuai Xu, Wenlu Zhang, Hailu Xu.

计算机科学 > 人工智能 arXiv:2610.22373 (cs) [提交于 2026 年 7 月 27 日] 标题：结构化权威注入：通过类型化 MCP 覆盖实现 LLM 代理的持久化人类控制 作者：Ryan Tran, YueMing Zhang, Lingfeng Tao, Nan Yang, Shengjie Xu, Yu Yin, Shuai Xu, Wenlu Zhang, Hailu Xu。

Abstract: LLM-based agents are increasingly used as autonomous orchestrators in distributed and multi-agent systems, yet a control gap remains: no protocol-level mechanism allows human operators to inject binding directives that persist across planning cycles. Free-text overrides suffer from ambiguous interpretation due to lack of schema validation and from authority erosion as operator instructions lose influence within accumulating context.

摘要：基于大语言模型（LLM）的代理正越来越多地被用作分布式和多代理系统中的自主编排器，但控制缺口依然存在：目前尚无协议层面的机制允许人类操作员注入在规划周期内持续有效的约束性指令。自由文本覆盖由于缺乏模式验证，容易导致解释歧义，且随着操作员指令在累积的上下文中逐渐失去影响力，会出现权威性削弱的问题。

We propose Structured Authority Injection (SAI), a protocol that addresses these issues by routing operator corrections through typed, schema-validated Model Context Protocol (MCP) tool calls and maintaining a ranked authority context prepended to each LLM prompt. SAI defines three override classes (Constraint, Priority, and Strategy) with distinct persistence semantics and an explicit precedence order enforced via a directed authority channel inaccessible to the planning process.

我们提出了结构化权威注入（SAI），这是一种通过类型化、经模式验证的模型上下文协议（MCP）工具调用来路由操作员修正，并维护一个置于每个 LLM 提示词之前的分级权威上下文的协议，从而解决上述问题。SAI 定义了三种具有不同持久性语义的覆盖类别（约束、优先级和策略），并通过一个规划过程无法访问的定向权威通道强制执行明确的优先级顺序。

The method requires no model fine-tuning or access to model weights and applies to instruction-tuned LLMs supporting MCP tool use. Evaluation across nine LLMs and eight scenarios targeting agent failure modes shows reduced ambiguous directive interpretation and mitigated authority decay across planning iterations, with lower prompt overhead in the strongest models (up to 38% depending on model and scenario), though benefit varies substantially across models.

该方法无需模型微调或访问模型权重，适用于支持 MCP 工具使用的指令微调 LLM。针对代理故障模式在九个 LLM 和八个场景中的评估表明，该方法减少了指令解释的歧义，缓解了规划迭代中的权威衰减，并在最强模型中降低了提示词开销（根据模型和场景不同，最高可达 38%），尽管其收益在不同模型间存在显著差异。

SAI provides a general design pattern for agent systems requiring explicit and persistent human control.

SAI 为需要明确且持久的人类控制的代理系统提供了一种通用的设计模式。