# ConnectOnion 1.9.0a30 — never published

**Never published.** The tag's release CI failed on a concurrent-fetch test
([#2253](https://github.com/openonion/connectonion/issues/2253)). These changes
first became installable in [1.9.0a31](/releases/1.9.0a31). Do not pin a30.

## Full-cohort concurrent onboarding and investigation

The a30 tag proposed removing the small default caps on onboarding and CLI
investigations, so a first run could investigate the full mapped cohort in
parallel:

1. **Full-cohort onboarding by default**: `co rem init` and `co rem investigate` investigate all mapped candidate people, all mapped projects with evidence, and all mapped organizations by default instead of truncating to small sample caps (`--first-*` flags remain available for opt-in dry runs).
2. **10-worker concurrent investigation pool**: `co rem investigate` and onboarding page synthesis use up to 10 workers by default. The originally projected 10–25 minute full-cohort duration was not verified in a completed trial.
3. **Thread-safe mail client sessions**: Mailbox tools isolate API connections per thread (`threading.local()` for Google client sessions) to ensure high-throughput parallel fetching without socket buffer collisions.
4. **Sandboxed task mask isolation**: Parallel runner processes maintain private temporary task masks (`private_task_mask()`) preventing concurrent tasks from clearing active permissions or review staging areas.

Design rationale is documented in `docs/design-decisions/077-concurrent-onboarding-investigation.md`.

These changes apply when `co rem` investigates pages. Stable remains 1.8.10.