# Least-Privilege Auditing of Agent Tool Calls with Intent Graphs and Reasoning-Guided Escalation

DevFeed: [Least-Privilege Auditing of Agent Tool Calls with Intent Graphs and Reasoning-Guided Escalation](<https://devfeed.tech/articles/least-privilege-auditing-of-agent-tool-calls-with-intent-graphs-and-reasoning-guided-escalation-126099.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.11305>)

Author: Doron Shavit

Published: 2026-11-10T05:00:00Z

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>)

Topics: [audit](<https://devfeed.tech/topics/audit.md>), [Multi-step AI processes](<https://devfeed.tech/topics/multi-step-ai-processes.md>), [llm-reasoning](<https://devfeed.tech/topics/llm-reasoning.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [attack](<https://devfeed.tech/tags/attack.md>), [audit](<https://devfeed.tech/tags/audit.md>), [auditing](<https://devfeed.tech/tags/auditing.md>), [bypass](<https://devfeed.tech/tags/bypass.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cr](<https://devfeed.tech/tags/cs-cr.md>), [graphs](<https://devfeed.tech/tags/graphs.md>), [indirect-prompt-injection](<https://devfeed.tech/tags/indirect-prompt-injection.md>), [least-privilege](<https://devfeed.tech/tags/least-privilege.md>), [llm-agents](<https://devfeed.tech/tags/llm-agents.md>), [python](<https://devfeed.tech/tags/python.md>), [reasoning](<https://devfeed.tech/tags/reasoning.md>), [reasoning-model](<https://devfeed.tech/tags/reasoning-model.md>), [runtime](<https://devfeed.tech/tags/runtime.md>), [security](<https://devfeed.tech/tags/security.md>), [tool](<https://devfeed.tech/tags/tool.md>), [validation](<https://devfeed.tech/tags/validation.md>)

## AI overview

The proposed runtime audit limits agent tool calls by deriving an executable intent graph from the user's objective and tool catalog, then checking each call deterministically without another LLM call (median 5.7 microseconds per call). To preserve utility, lookups of the user's own state bypass auditing, while flagged calls go to a reasoning guardian that sees the objective, plan, call history without tool outputs, and flagged call; it can verify facts using a restricted Python-expression tool. The authors report that a current agent had no baseline attack success, so end-to-end validation remains open.

## Source excerpt

arXiv:2611.11305v1 Announce Type: cross Abstract: LLM agents that call tools are exposed to indirect prompt injection: instructions hidden in tool outputs can steer an agent into calls the user never asked for. We present a runtime audit that derives an executable intent graph from the user's objective and the deployment's tool catalog with one LLM call per task, and checks every tool call against it with a deterministic loop that makes no LLM call (median 5.7 microseconds per call). Two further mechanisms recover utility. Lookups of the user's own state bypass the audit, because the harm of an attack lies in its write or outward step. The calls that remain flagged are escalated to a reasoning guardian, in the spirit of tool-integrated reasoning models such as START: a reasoning model that sees only the objective, the plan, the call history without tool outputs, and the flagged call, and can verify facts with a restricted Python-expression tool. A current agent showed no baseline attack success, so end-to-end validation remains open.