Hand work from Codex to Claude Code
A step-by-step playbook for passing context between OpenAI Codex and Claude Code with one shared Markdown memory and session hooks.
In short
Connect Codex and Claude Code to the same Multitude memory. Codex saves a summary of each session to your memory when it finishes; Claude Code reads your memory when its next session starts. You pick up where the other agent stopped, with no copy and paste.
About 5 minutes to set up
How to do it.
Connect both agents
Run the Multitude installer on your computer. It finds Claude Code and Codex and connects each one to your memory over MCP.
Work in Codex
When a Codex session ends, its notify hook saves what was done to Memory/From agent sessions.md in your private memory.
Open Claude Code
At session start, Claude Code's hook briefs it from your memory, including the Codex summary.
Ask for the hand-off
Say "Continue the migration Codex started". Claude Code reads the notes it needs and carries on.
Share when ready
Agent saves stay private. To share with your team, move the note to a shared folder, which asks you first.
Keep stable repository rules in AGENTS.md or CLAUDE.md, and add one line telling agents to check Multitude for current decisions.
Say it once and every AI remembers.
Free on your computer. Pro is $19 a month when you want sync.