HybridClaw 0.36.0 adds shared preferences and work records
The assistant runtime connects explicit feedback with scheduled work
HybridClaw released version 0.36.0 on October 4, adding shared preferences and records explaining background suggestions to its AI assistant runtime.
Preferences across chats and schedules
One record per authenticated user carries explicit feedback into chats, companion apps and scheduled generation. Existing schedules read changes on their next run. Reading or bookmarking alone does not create a preference.
What the records can show
Scheduled work can save reasons, sources, artifacts and action receipts. The work records distinguish completion, chat storage, phone notification attempts and viewing. Older results do not receive retrospective evidence. Saved references are not independently verified claims.
New memory views group information by people, groups and sessions. Journals retain memory changes, including facts later removed from current memory.
Deploy the gateway and rebuilt worker together before enabling the companion app features, and retain the runtime database. The preferences guide reports no live model compliance testing.
For teams evaluating persistent assistants, test a changed preference in an existing schedule and inspect the resulting record. Check whether the explanation actually supports the suggestion before relying on it.
ByteForward reviewed release notes and versioned documentation. We have not installed or tested HybridClaw.
Archival phone photograph by Dennis Cortés / CC0, published October 9, 2016. Used unchanged as an illustration. This is not a HybridClaw app demonstration.
