Matoi: Selective Revalidation of Consequence Evidence for Agent Authorization
Matoi reuses checked consequence evidence for agent authorization, cutting lifecycle cost a few percent in tests.
Matoi separates reusable consequence evidence from current execution authority and durable occurrence history under a finite query-aware contract. In a 1,440-cell matched-wire study with identical compact inputs, reuse reduced complete lifecycle cost by 3.49 percent in JavaScript and 1.79 percent in Python; combined representation and reuse cuts versus full proof were 10.87 and 16.53 percent. A frozen AgentDojo Banking mapping covered 5 of 16 user plans and 7 of 9 attack plans, preserving five exact benign transfers and refusing 35 reused attack-plan and policy crosses. The authors say the tools are scripted synthetics, not autonomous agents or a real bank, and do not claim production superiority.
- Reuse cut lifecycle cost 3.49% in JavaScript and 1.79% in Python.
- Combined representation and reuse savings reached 10.87% and 16.53%.
- AgentDojo mapping refused 35 reused attack-plan and policy crosses.
- Evaluation uses scripted synthetic tools, not a real bank or autonomous agents.
From the original article
1 min read at arxiv.orgMatoi separates reusable consequence evidence from current execution authority and durable occurrence history. Its finite query-aware contract covers branches, ordered collections and negative lookups. A consumer reuses only its own checked evidence whose observations remain valid; signed requests, live grants, quota and uncertain outcomes remain separate obligations.
By Zexun Wang · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.