Answers are grounded in this paper and cite their sources.
AI agents accumulate rich episodic context events, decisions, roles, intentions but that context is trapped in vendor silos. No standard format exists for transporting, storing, and sharing episodic memory across agents, devices, and organizational perimeters with pri- vacy guarantees and granular access control. We introduce the AIX Protocol, an open specification for portable episodic identity memory comprising three integrated components: (1) the .aix envelope, a portable JSON format carrying a 9-dimensional identity snap- shot, spatiotemporal provenance, and per-field permission declarations; (2) Nexus v0.1, a permission and federation layer governing which memory dimensions are shared between which perimeters (self, organization, agent) under which role conditions; and (3) a privacy layer operating before transmission under Patent Application P202631047 (OEPM, filed July 19, 2026), providing pre-inference transformation, intra-session coherence, and inter-session opacity. The Knowledge Compiler preprocessing pipeline reduces context token consump- tion by 35-78% versus raw text injection through deduplication and JSON distillation; the token_budget field enables receiving agents to plan context usage before loading an enve- lope. Merkle-SHA256 integrity signatures over the memory array enable tamper detection. We describe the reference implementation aix_parser.py (Python standard library only, Apache 2.0, 23 tests), Self v1.1 as the frst production implementation across 11 passive in- gestion sources, and the SOMA enterprise deployment pattern. The specification is available at https://github.com/alanlapid/aix-protocol (v0.2.0, August 2026).
No takes yet. Share an insight, caveat, or question.
Alan Lapid (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: