Beyond Approved Actions: Runtime Validation of Persistent Outcomes in Agent Workflows
cs.SE, cs.AI
Submitted: 2026-09-25
Updated: 2026-09-25
Comments: 22 pages including 7 pages of supplementary material. Submitted to IEEE Transactions on Software Engineering
Code: https://github.com/eval-sys/mcpmark2https:
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- AIRGuard: Guarding Agent Actions with Runtime Authority Control
- SecureClaw: Clawing Back Control of LLM Agents
- Temporary Authority, Permanent Effects: Commit-Time Authorization for LLM Agents
- Beyond Single-Use Tokens: Durable Authorization State for Replay-Resistant LLM Agent Actions
- MCPMark: A Benchmark for Stress-Testing Realistic and Comprehensive MCP Use
- Defeating Prompt Injections by Design
- Securing AI Agents with Information-Flow Control
- Cordon: Semantic Transactions for Tool-Using LLM Agents
- Atomix: Timely, Transactional Tool Use for Reliable Agentic Workflows
- Progent: Securing AI Agents with Privilege Control
- What You Approve Is What Executes: Consent Integrity for Black-Box LLM Agents
- Proof-Carrying Agent Actions: Model-Agnostic Runtime Governance for Heterogeneous Agent Systems
- Verified Tool Calls Improve LLM Agent Reliability Under Non-Atomic Failures
Related papers
- Falsification-Based Verification of LLM-Generated Optimization Models: Sound Test Batteries and Their Detection Limits
- GitSkills: A Dataset of Agent Skills on GitHub
- SABER: Benchmarking Operational Safety of LLM Coding Agents in Stateful Project Workspaces
- PackMonitor: Enabling Zero Package Hallucinations Through Decoding-Time Monitoring
- IntentCoding: Amplifying User Intent in Code Generation
- Incentives and Outcomes in Bug Bounties