Preprint / scholarly article

When Should a Local Agent Act, Assist, Verify, Withdraw, or Exit? A Certified Local Micro-Theory of Open-Task Participation

K. Takahashi

Published
DOI
10.5281/zenodo.19394600

Full text PDF (Zenodo)

Abstract

This preprint develops a certified local micro-theory of open-task participation in agent societies. It formalizes when an authenticated local agent should act, assist, verify, withdraw, or exit under public evidence, certified uncertainty, attribution, and implementability constraints.

Keywords

  • multi-agent systems
  • local agent participation
  • open-task participation
  • decentralized decision making
  • auditable AI
  • agent verification
  • open-world agents
  • human-AI coordination
  • verifier portfolios
  • authenticated snapshots
  • certified uncertainty
  • agentic AI
  • task allocation
  • decentralized control
  • local micro-theory
  • participation governance

Identifiers and source records