SuperLocalMemory 4.0: The Governed Memory Operating System for AI Agents

arXiv:2608.08253v2 Announce Type: replace Abstract: We present SuperLocalMemory 4.0, a governed, local-first memory operating system for AI agents, unifying multi-channel retrieval under reciprocal-rank fusion, bi-temporal recall, multi-scope isolation, role-based access, verified erasure, and a hash-chained audit trail. A reliability spine governs the primary write path: generation-fenced admission, verifiable memory transactions with per-projection apply, verify, compensate and erase owners, and hash-checkable completion manifests. Eleven fault-injection scenarios, each repeated 200 times, upheld 2,199 of 2,200 scoped component properties. This version leads with a negative result. Ten mechanisms here were implemented, reachable on a live call path, and ineffective at their final connection. Implemented, reachable and effective are three different questions, and the third requires an oracle independent of the mechanism under test. We contribute two mechanical invariants that supply one: a prior-distance assertion over Bayesian learners, and a join-liveness assertion over schema-guarded paths that reports where a guard's missing data resides. A three-arm ablation varying only the recall session-identifier namespace moves no posterior with the defect present and every instantiated arm with it absent, while a negative control that writes every ticket but supplies no engagement settles nothing. We withdraw the previous version's governed write-envelope overhead figure: the two paths it differenced are not comparable. Timing the envelope in place gives an 11.0 ms governed write of which the envelope is 70.6 percent, but the generation fence costs 1.9 microseconds and the obligation ledger 42 microseconds. The cost is durability, not governance.

Abstract:We present SuperLocalMemory 4.0, a governed, local-first memory operating system for AI agents, unifying multi-channel retrieval under reciprocal-rank fusion, bi-temporal recall, multi-scope isolation, role-based access, verified erasure, and a hash-chained audit trail. A reliability spine governs the primary write path: generation-fenced admission, verifiable memory transactions with per-projection apply, verify, compensate and erase owners, and hash-checkable completion manifests. Eleven fault-injection scenarios, each repeated 200 times, upheld 2,199 of 2,200 scoped component properties. This version leads with a negative result. Ten mechanisms here were implemented, reachable on a live call path, and ineffective at their final connection. Implemented, reachable and effective are three different questions, and the third requires an oracle independent of the mechanism under test. We contribute two mechanical invariants that supply one: a prior-distance assertion over Bayesian learners, and a join-liveness assertion over schema-guarded paths that reports where a guard's missing data resides. A three-arm ablation varying only the recall session-identifier namespace moves no posterior with the defect present and every instantiated arm with it absent, while a negative control that writes every ticket but supplies no engagement settles nothing. We withdraw the previous version's governed write-envelope overhead figure: the two paths it differenced are not comparable. Timing the envelope in place gives an 11.0 ms governed write of which the envelope is 70.6 percent, but the generation fence costs 1.9 microseconds and the obligation ledger 42 microseconds. The cost is durability, not governance.

54 pages, 15 figures, 8 tables. Substantially revised: retracts two claims from v1 and adds two controlled experiments. Code: this https URL Zenodo DOI: https://doi.org/10.5281/zenodo.21853302. Code: this https URL

Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)

arXiv:2608.08253 [cs.AI]

(or arXiv:2608.08253v2 [cs.AI] for this version)

aiqbench analysis

SuperLocalMemory 4.0: The Governed Memory Operating System for AI Agents

Related coverage