# ConfGuard: a physics-grounded reference monitor for predictive-maintenance large-language-model agents

DevFeed: [ConfGuard: a physics-grounded reference monitor for predictive-maintenance large-language-model agents](<https://devfeed.tech/articles/confguard-a-physics-grounded-reference-monitor-for-predictive-maintenance-large-language-model-agents-140002.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.22288>)

Author: Arunkumar V, Gangadharan G. R., S. Senthilkumar

Published: 2026-11-18T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Cryptography and Security](<https://devfeed.tech/sources/arxiv-cryptography-and-security.md>), [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [LLM security](<https://devfeed.tech/topics/llm-security.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [LLM evaluation / benchmarking](<https://devfeed.tech/topics/llm-evaluation-benchmarking.md>), [Retrieval Augmented Generation (RAG)](<https://devfeed.tech/topics/retrieval-augmented-generation-rag.md>)

Tags: [adversarial](<https://devfeed.tech/tags/adversarial.md>), [agents](<https://devfeed.tech/tags/agents.md>), [attack](<https://devfeed.tech/tags/attack.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cr](<https://devfeed.tech/tags/cs-cr.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [grounded](<https://devfeed.tech/tags/grounded.md>), [large-language-model](<https://devfeed.tech/tags/large-language-model.md>), [maintenance](<https://devfeed.tech/tags/maintenance.md>), [monitor](<https://devfeed.tech/tags/monitor.md>), [physics](<https://devfeed.tech/tags/physics.md>), [predictive-maintenance](<https://devfeed.tech/tags/predictive-maintenance.md>), [qwen3](<https://devfeed.tech/tags/qwen3.md>), [reference](<https://devfeed.tech/tags/reference.md>), [sensor](<https://devfeed.tech/tags/sensor.md>)

## AI overview

ConfGuard combines a history-anchored conformal monitor for remaining useful life with a validator that checks maintenance urgency against a deterministic policy. Its physics rule caps remaining life, which must never increase and may fall by at most one cycle per cycle; trajectory-level calibration lets operators target an expected escalation share for clean windows. On 10,196 C-MAPSS test windows, it reduced covert attacks on healthy assets from 0.529 to 0.046 and failing-asset suppression from 0.151 to 0, escalating 0.027 of clean windows against a 0.05 target. With a Qwen3-14B planner, the combined monitors reduced urgency-class injection to 0.00 and universal sensor attack success to 0.12, versus 0.29 and 0.54 for a defended prompt. The abstract also reports an attack class the projection misses.

## Source excerpt

arXiv:2611.22288v1 Announce Type: new Abstract: Large-language-model agents now close the industrial predictive-maintenance loop, reading sensor windows, retrieving procedures and dispatching work orders. Two attack channels meet there: injected text in a retrieved document can hijack the decision, and a bounded sensor perturbation can corrupt the remaining-useful-life estimate behind it. Recent agent defenses police data provenance and tool privilege but cannot tell whether a prognostic quantity is physically plausible. ConfGuard is a reference monitor with a physics-grounded policy. Its conformal monitor tests each prediction, anchored to the asset's admitted history, against degradation physics: capped remaining life never rises and falls at most one cycle per cycle. Calibrating its threshold over whole asset trajectories lets the operator set the expected share of a clean asset's windows the running monitor escalates. A physics-policy validator checks the agent's maintenance-urgency decision against the decision a deterministic policy derives from the same interval. On all 10,196 C-MAPSS test windows the monitor cuts a covert attacker's reach on healthy assets from 0.529 to 0.046 and closes suppression of failing assets (0.151 to 0), escalating 0.027 of clean windows against a 0.05 target, while the fixed physical range it replaces leaves both attacks open. Behind a Qwen3-14B planner the composed monitors hold urgency-class injection to 0.00 and the universal sensor attack to 0.12, against 0.29 and 0.54 for a defended prompt. An adaptive attacker defeats a Mahalanobis detector on 0.504 of windows and adversarial training on 0.485. We report an attack class the projection misses.