COMPARE / RESUME
Resume on this Mac vs a searchable archive
Resume is built into Claude Code and Codex: it reopens one saved session on the Mac where that session lives. A searchable archive uploads masked transcripts so any MCP agent can query every Mac in your workspace—not replace resume, but cover the gaps resume leaves.
What resume does well
Resume is fast and offline. You already know the project folder; claude --continue or codex resume --last puts you back in context with tools and files as you left them. No account, no upload—just local JSONL or session stores.
Where resume stops
- Another Mac — history on the laptop does not appear in the desktop picker.
- Another agent — Cursor cannot resume a Codex thread; each tool reads its own files.
- Fuzzy recall — you remember the bug, not the session id or repo path.
- Counts and audits — resume does not answer “how many chats mentioned migration?”
What an archive adds
AIChatMemory masks Claude Code, Codex, and Cursor chats on each Mac, uploads them to your workspace, and serves MCP tools search, count, and session. Agents retrieve snippets or full masked chats and answer in their own words. Masking is not a guarantee that no sensitive text remains.
Use both
Keep using resume for day-to-day continuity in the repo you have open. Add the archive when you need shared, cross-machine memory of what your agents already said—in code, in terminals, and in IDE threads.
See also: Resume cheat sheet · MCP memory vs archive MCP · All guides