NEXO Brain is not only a memory plugin. It is a local cognitive runtime that gives multiple clients the same brain, keeps operational state durable, and exposes enough surface area to be audited instead of hand-waved.
One runtime, one local brain, multiple clients. Context and operator state survive when work moves from Claude Code to Codex or to any MCP surface wired into the same runtime.
NEXO stores memory locally, can search it semantically, and keeps enough operational evidence that the product claim can be audited instead of trusted blindly.
Tasks, checkpoints, closeout evidence, reminders, and followups let long-running work survive compaction, restarts, and overnight analysis.
Preferences, automations, email accounts, model-quality tiers, and support surfaces are becoming product controls instead of local tribal knowledge.
nexo doctor, startup preflight, fail-closed checks, runtime audits, and guarded updates reduce silent drift and make failures diagnosable.
Website, docs, changelog, scorecards, blog, and release validators aim to describe the product as it really exists, not as a stale narrative.
Short product narratives: install path, shared brain across clients, evolution loop, and why NEXO is more than memory.
Where NEXO fits for Claude Code, Codex, MCP deployments, and teams that want a local runtime instead of a hosted memory layer.
The review-gated self-improvement loop that turns real runtime usage into bounded, auditable change proposals.