Capture pipeline

When you run profile upload, profile refresh, or profile inspect, the CLI executes this pipeline:

1. Copy — A filtered snapshot of your local browser profile (skipping caches, GPU data, and lock files) is written to a temp directory.
2. Launch — Your real browser binary is spawned headless against the temp copy with sync, telemetry, and Safe Browsing disabled so nothing mutates your live profile.
3. Extract — The CLI connects via CDP (Storage.getCookies + per-origin localStorage / IndexedDB) and captures all auth state.
4. Filter — Domain exclusions (persistent list + per-run --exclude-domain) and the private-domain filter are applied. Excluded data never leaves your machine.
5. Upload — The filtered payload is sent to the Browserless server.
6. Cleanup — The temp directory and headless process are torn down.

Profile size limits

The server enforces these caps on the upload payload:

Limit	Value
Serialized artifact size	2 MB
Origins per capture	50
IndexedDB databases per origin	5
IndexedDB entries per store	1,000
Profile name length	255 characters
Profile name pattern	No whitespace, ?, or #

If your capture exceeds 2 MB, use --auto-fit to drop the heaviest origins until it fits, or --exclude-domain / --only-domain to narrow scope.