A new method for reusing evidence in agent authorization

Matoi separates reusable evidence from current authority to reduce lifecycle costs.

Researchers developed Matoi for selective revalidation of consequence evidence. It separates reusable consequence evidence from current execution authority and durable occurrence history.

The system uses a finite query-aware contract. This contract covers branches, ordered collections, and negative lookups. A consumer reuses only its own checked evidence where observations remain valid.

Signed requests, live grants, quota, and uncertain outcomes remain separate obligations. Conditional correctness arguments distinguish locally serialized claims from nontransactional remote effects. Pre-satisfied utility predicates are reported separately from completed work.

A 1,440-cell matched-wire study repeated 144 authored inputs. With compact input bytes identical, reuse reduces complete lifecycle cost by 3.493.49 percent in JS and 1.791.79 percent in Python. Combined representation/reuse reductions against full proof are 10.8710.87 and 16.5316.53 percent.

An AgentDojo Banking mapping covers 5/165/16 user plans and 7/97/9 attack plans. The strategies preserve five exact benign transfers and refuse 35 reused attack-plan/policy crosses. These were original scripted synthetic tools, not autonomous agents or a real bank.

The results support a bounded, auditable reuse and execution contract. This is not a new general incremental algorithm, arbitrary-tool closure, atomic third-party commitment, or production superiority.

Key facts

  • Reuse reduces complete lifecycle cost by 3.493.49 percent in JS and 1.791.79 percent in Python.
  • The AgentDojo Banking mapping covers 5/165/16 user plans and 7/97/9 attack plans.
  • The system refuses 35 reused attack-plan/policy crosses.

All sections

Edition