SiderMem
Back to Home

How Can Codex and Claude Code Share Memory?

Capture a memory or context file from one coding agent's session with sidermem-cli, then search or read it back from the other agent's terminal. The memory is cloud-synced and available from any terminal you're logged into, independent of either tool.

The problem

Coding agents typically keep context local to their own session or memory file format — a decision made in a Codex session doesn't automatically show up when you switch to Claude Code on the same project, or on a different machine.

Capture from the first agent's session

sidermem create --file ./AGENTS.md -t "Codex session notes"

Or pipe a summary of what happened directly:

git diff main | sidermem create -t "Codex — auth refactor context"

Read it back from the second agent's terminal

sidermem list --search "auth refactor"
sidermem show <id>

Both commands work identically whether you're now in a Claude Code session, a plain terminal, or a different machine entirely — the CLI reads from your cloud-synced memory, not from either agent's local state.

Safety notes

Anything captured this way is uploaded once synced — review agent-generated context and diffs for secrets or credentials before capturing, same as you would before committing to a shared repository.

Related: CLI overview · Piping terminal output into your knowledge base

View on NPM