arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2603.18829cs.CRcs.AI
收藏

代理控制协议:代理行为的准入控制

Agent Control Protocol: Admission Control for Agent Actions

TraslaIA

Marcelo Fernandez

AI总结:

本文提出ACP协议,通过静态风险评分与状态信号结合,控制代理行为,实验显示其能显著降低恶意行为执行率,同时通过形式化验证确保安全性和活性。

摘要:

自主代理可能产生有害行为模式,因每个请求独立评估无法解决。本文提出ACP协议,通过静态风险评分与状态信号结合,通过LedgerQuerier抽象实现行为属性的控制。在500请求负载下,无状态引擎批准所有请求,而ACP仅允许2/500执行(0.4%),并识别ACP-RISK-2.0中的状态混合漏洞,引入ACP-RISK-3.0改进异常信号范围。决策评估耗时739-832纳秒,吞吐量达172万请求/秒。安全性和活性通过TLA+验证,形式化定义偏差崩溃并引入边界激活率(BAR)作为检测机制。协调代理独立积累风险,协调窗口CW_appr=2N无偏差,活动线性增长,防止超线性放大。ACP是Agent Governance系列论文之一。

英文摘要:

Autonomous agents can produce harmful behavioral patterns from individually valid requests -- a threat class per-request policy evaluation cannot address, because stateless engines evaluate each request in isolation. We present ACP, a temporal admission control protocol enforcing behavioral properties over execution traces via static risk scoring combined with stateful signals (anomaly accumulation, cooldown) through a LedgerQuerier abstraction. ACP blocks execution based on deterministic, history-aware risk scoring -- not anomaly detection. Under a 500-request workload where every request is individually valid (RS=35), a stateless engine approves all 500; ACP limits autonomous execution to 2 out of 500 (0.4%), escalating after 3 actions and denying after 11. We identify a state-mixing vulnerability in ACP-RISK-2.0 (cross-context false denials) and introduce ACP-RISK-3.0, scoping anomaly signals to PatternKey(agentID, capability, resource). Decision evaluation: 739-832 ns (p50); throughput 1,720,000 req/s. Safety and liveness model-checked via TLA+ (11 invariants + 4 temporal properties, 0 violations) across 4,294,930,695 distinct states. We formalize deviation collapse -- enforcement active but never exercised due to upstream constraints -- and introduce Boundary Activation Rate (BAR) as its detection mechanism. An adversary suppressing BAR to 0.00 is detected via DeltaBAR before collapse (BAR_C=1.00). N coordinated agents accumulate risk independently; coordination window CW_appr=2N with zero deviation: activity scales linearly, preventing superlinear amplification. ACP is Paper 1 of a 6-paper Agent Governance Series: P0 -- atomic decision boundaries; P2 -- behavioral drift detection (IML); P3/4 -- governance structure, fair allocation, and irreducibility; P5 -- runtime execution validity (RAM, arXiv:2604.22898); P6 -- operationalization of RAM.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑