The key thing to understand: Nessie imports data from the machine or service that can actually see it, then Cloud Sync makes the imported Nessie data available elsewhere.
For an overview of what Cloud Sync uploads and what stays local, see What Cloud Sync does.
Cloud Sync does not copy your provider login sessions, browser cookies, SSH credentials, or local file access to other machines. It copies the Nessie library records created from those sources.
Mental model
- A local importer reads the source it has access to, such as a browser account, a local Codex folder, an Obsidian vault, or an OpenClaw host
- Nessie stores the imported transcripts, notes, contexts, profile updates, and source graph metadata in the local library
- If Cloud Sync is enabled, those library records sync to Nessie Cloud
- Other surfaces can use the cloud copy, or pull cloud records into their own local library
Current access paths
| Surface | Data source | What that means |
|---|---|---|
| Nessie app on your Mac | Local library, plus records pulled down through Cloud Sync | The app can show your local imports and synced cloud records. New teammate-shared data may need a sync pull before it appears locally. |
| Nessie CLI on your Mac | The running Nessie app over localhost | Today the CLI uses the local app, so the app must be running and current. |
| MCP clients | Nessie Cloud | Claude, ChatGPT, OpenClaw, and similar MCP clients use the hosted cloud copy instead of your Mac's local database. |
| Remote daemon or cloud worker | Its own local files plus Nessie Cloud through an API key | The remote machine imports what it can see, then syncs those records into your Nessie library. See Nessie daemon. |
| Another Mac | Its own local library plus Cloud Sync | The second Mac receives synced Nessie records, but it does not receive your provider credentials or local filesystem access. |
| Team-shared sources | The source owner's synced records | A teammate chooses what to share. Those records become available through Cloud Sync with owner and team provenance. |
Examples
- Codex on your Mac: Nessie reads local Codex transcripts on that Mac. The local CLI can search them through the running app. With Cloud Sync on, MCP clients and teammates can use the synced copy.
- Claude using MCP: Claude cannot read your Mac's local database. It searches the Cloud Sync copy, so Cloud Sync must be enabled and up to date.
- A remote Linux or cloud agent: Install the Nessie daemon, configure it with an API key, and let that remote machine import its own agent sessions. You can reuse one API key for a group of short-lived workers when you want them grouped under one source.
- A teammate's shared session: Their machine or daemon syncs the session to Nessie Cloud. You can then find it through MCP or CLI/team workflows, depending on which surface has the latest synced view.
Freshness
Already-synced records stay available from cloud-backed surfaces even if the original Mac is offline. New local files, new browser conversations, and new remote agent sessions appear after the importer that can see them runs and syncs.
The CLI is local-first today. Cloud-backed CLI paths for users who want direct cloud search and faster shared-source access are planned, while local app access remains useful for Mac-local workflows.