iORBi (Intent-Oriented Reasoning Bridge Interface) is a structured language framework designed to enable precise, auditable, and semantically grounded coordination between human principals and autonomous AI agents. The framework introduces a formal grammar for intent expression, decomposing high-level directives into verifiable action primitives that AI systems can execute, validate, and trace back to originating human intent. iORBi addresses a fundamental gap in current human-AI interfaces: the absence of a shared, machine-readable substrate for intent that is simultaneously legible to human collaborators. The framework defines five canonical intent layers — Directive, Constraint, Context, Outcome, and Attestation — and specifies canonical serialisation in both JSON and a compact binary codec (the 5-dip codec introduced by the Swiss Vault IP specification). iORBi is designed to operate within sovereign runtime environments such as Pyraclaw OS and the iAiA multi-agent swarm topology, and its attestation layer is compatible with ORCID-linked research identity records and Zenodo DOI anchoring. A reference implementation is provided for the WISeer 77-node Claude Agent SDK topology, demonstrating round-trip intent serialisation, execution, and cryptographic attestation using SHA3-512 capsule sealing. The framework is offered as a technical note to invite community feedback ahead of a full specification release.
Callaghan et al. (Tue,) studied this question.