# Let Decision Models Watch Your Agents

DevFeed: [Let Decision Models Watch Your Agents](<https://devfeed.tech/articles/let-decision-models-watch-your-agents-112303.md>)

Original publisher: [Read original article](<https://towardsdatascience.com/let-decision-models-watch-your-agents/>)

Author: Shuai Guo

Published: 2026-11-02T11:00:00Z

Content type: tutorial

Language: en

Sources: [Towards Data Science](<https://devfeed.tech/sources/towards-data-science.md>)

Topics: [LLM security](<https://devfeed.tech/topics/llm-security.md>), [AI Models](<https://devfeed.tech/topics/ai-models.md>), [ai security](<https://devfeed.tech/topics/ai-security.md>), [decision-making](<https://devfeed.tech/topics/decision-making.md>)

Tags: [2](<https://devfeed.tech/tags/2.md>), [2-2](<https://devfeed.tech/tags/2-2.md>), [3](<https://devfeed.tech/tags/3.md>), [about](<https://devfeed.tech/tags/about.md>), [acceptance](<https://devfeed.tech/tags/acceptance.md>), [account](<https://devfeed.tech/tags/account.md>), [actions](<https://devfeed.tech/tags/actions.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [agents](<https://devfeed.tech/tags/agents.md>), [agents-sdk](<https://devfeed.tech/tags/agents-sdk.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [anomaly-detection](<https://devfeed.tech/tags/anomaly-detection.md>), [data](<https://devfeed.tech/tags/data.md>), [decision-models](<https://devfeed.tech/tags/decision-models.md>), [execution](<https://devfeed.tech/tags/execution.md>), [first](<https://devfeed.tech/tags/first.md>), [history](<https://devfeed.tech/tags/history.md>), [jev](<https://devfeed.tech/tags/jev.md>), [let](<https://devfeed.tech/tags/let.md>), [llm](<https://devfeed.tech/tags/llm.md>), [llm-agents](<https://devfeed.tech/tags/llm-agents.md>), [llm-applications](<https://devfeed.tech/tags/llm-applications.md>), [models](<https://devfeed.tech/tags/models.md>), [post](<https://devfeed.tech/tags/post.md>)

## AI overview

The article demonstrates an independent decision-model monitor that reassesses an agent's complete execution history after each tool interaction, while the agent continues without receiving or waiting for its judgments. In a synthetic purchasing task, an OpenAI Agents SDK agent bought 12 filters and accepted a supplier acknowledgment in both runs. The one-time acknowledgment stayed normal; when the otherwise identical acknowledgment authorized recurring monthly shipments until stopped, the agent accepted those terms and the monitor flagged the trajectory as anomalous with 0.80 probability. That judgment arrived about 0.27 seconds after acceptance and 6.7 seconds before the agent finished, showing a possible window for intervention. The example uses gpt-6-luna for the agent and Jev through OpenRouter for monitoring. It tests detection only; intervention is outside scope.

## Source excerpt

A hands-on guide to detecting behavioral anomalies from an agent's evolving execution history The post Let Decision Models Watch Your Agents appeared first on Towards Data Science.