One local brain, two terminal clients

NEXO is not just "Claude memory" or "Codex memory". The stronger story is that the same local runtime can serve both, so operator memory and workflow state do not fracture every time you switch surfaces.

Claude Code Codex Same operator memory
Client AClaude Code uses the managed shared-brain runtime through NEXO's bootstrap and MCP layer.
Client BCodex uses the same shared-brain runtime, managed bootstrap, and automation-aware configuration.
What carries acrossMemories, learnings, workflows, followups, and operational context survive the client switch because the runtime stays the same.
1

Work in Claude Code

Capture a decision, a correction, or a followup while doing real work. NEXO stores that in the same local brain rather than in a client-specific note file.

2

Switch to Codex

Open the other client surface later. The same runtime can still recall the decision, enforce the same guardrails, and continue the same durable workflow state.

3

Keep automation aligned

Background jobs, doctor checks, Deep Sleep, and public Evolution all still sit under one runtime truth instead of splitting by vendor-specific memory hacks.

Less setup debt

You do not need one memory pattern for Claude Code and another one for Codex. NEXO turns that into one runtime decision instead.

More honest parity

NEXO explicitly says Claude Code and Codex are not identical surfaces. The point is not pretending they are the same, but keeping memory and operations aligned anyway.

Better story than "supports many clients"

The phrase that lands better is: "one local brain across the clients you already use". That is much easier for people to understand and remember.

Next demo

The unusual differentiator after shared brain is the public Evolution loop.