Sessions, history & cloud sync
Where your conversations live, what syncs between machines, and how a dropped connection never loses an agent run.
Saved sessions
Every conversation is a session you can come back to — full chat history, rendered Markdown and all.
| Plan | Saved sessions |
|---|---|
| Standard | One saved session at a time |
| Pro | Multiple saved sessions |
| Teams | Multiple, shared tooling across the org |
What syncs to the cloud
On Standard and above, your setup follows you across machines:
- Chat sessions — pick up a conversation on another box.
- Tool preferences — which agent tools are enabled and how they're scoped.
- Custom personas (Pro and Teams) — personas you define sync with you; Teams shares them org-wide.
Your provider API keys are stored encrypted at rest and never shared across accounts — see Providers & BYOK for how key storage works.
Runs that survive a dropped connection
The multi-turn agent loop executes server-side: when the model works through a task — reading files, planning, editing — that loop keeps running even if your IDE disconnects. Tool execution (the actual reads, edits, and commands) still happens locally in your IDE; reconnect and the run picks back up mid-turn instead of starting over.
The same sessions in the CLI
CLI runs are saved automatically too. List and resume them from any terminal:
tempr history
tempr --id <SESSION_ID> "continue where we left off"