Authorization Closure Graph: Minimal Repair for LLM Agents with Evolving User Instructions

arXiv:2609.32428 · cs.AI · Submitted 2026-09-26 · Read on arXiv

cs.AI

Submitted: 2026-09-26

Updated: 2026-09-26

Comments: 22 pages, 8 figures, 6 tables. Preprint under review

Code: https://github.com/weiliang822/ACG

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Benchmark — 11× in this paper · explained in 2 episodes
• Ablation — 9× in this paper · explained in 1 episode
• Authorization Closure Graph — 7× in this paper
• Action Safety — 5× in this paper
• Actor — 5× in this paper · explained in 3 episodes
• Agentic Workflows — 2× in this paper · explained in 1 episode
• LLM judge — 2× in this paper · explained in 2 episodes
• Provenance — 2× in this paper · explained in 3 episodes
• Topological Order — 2× in this paper · explained in 2 episodes
• AI Agents — 1× in this paper · explained in 2 episodes
• Ablation Studies — 1× in this paper · explained in 1 episode
• Abstraction — 1× in this paper · explained in 1 episode
• Agent Skills — 1× in this paper · explained in 1 episode
• Agentic System — 1× in this paper · explained in 2 episodes
• Large Language Model (LLM) — 1× in this paper · explained in 3 episodes
• Trace — 1× in this paper · explained in 2 episodes
• Verifier — 1× in this paper · explained in 2 episodes

Sources

• $\tau^2$-Bench: Evaluating Conversational Agents in a Dual-Control Environment
• Securing AI Agents with Information-Flow Control
• AgentDojo: A Dynamic Environment to Evaluate Prompt Injection Attacks and Defenses for LLM Agents
• Defeating Prompt Injections by Design
• PolicyGuard: A Dialogue-Grounded Sub-Agent Verifier for Policy Adherence in LLM Agents
• VIGIL: Runtime Enforcement of Behavioral Specifications in AI Agent Skills
• ToolSandbox: A Stateful, Conversational, Interactive Evaluation Benchmark for LLM Tool Use Capabilities
• Formal Policy Enforcement for Agentic Systems
• Near-Miss: Latent Policy Failure Detection in Agentic Workflows
• Temporary Authority, Permanent Effects: Commit-Time Authorization for LLM Agents
• Progent: Securing AI Agents with Privilege Control
• Aligning Provenance with Authorization: A Dual-Graph Defense for LLM Agents
• What You Approve Is What Executes: Consent Integrity for Black-Box LLM Agents
• Beyond Single-Use Tokens: Durable Authorization State for Replay-Resistant LLM Agent Actions
• FAVA: Formal Authorization for Verified Agents with Evidence-Backed Permission Graphs

Related papers

• MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
• Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
• Agents' Last Exam
• The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
• MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
• Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems