[kinqsradio]

Summary

• Background crews and their runs are stored per device in the data folder (schema 41), not in .orbyn/crew.json (rule 4).
• A crew: template (fix failing CI, flaky test, dependency bump, security advisory), trigger (CI failure, issue label, schedule, dependency update) with strict settings, a run budget fixed at creation, an enabled flag; created off.
• Runs track state (running, stopped with reason, needs review) and spend against budget. Reaching the budget stops the run with reason budget and queues it for review. No merge authority.
• Routes under /api/crew/background: list, read, runs; create, update, enable, disable, delete. Preview-gated, local only for writes, listed in the N1 marker. Every change writes an audit row in the same transaction.
• Not built: triggers, any engine start (slice 4h), UI, joining a budget-stopped run to the after-merge Review queue.

Closes #

none (no issue)

Parity rows

none

Frames built

Frames built: none

Testing

• CI covers the rest.
• Locally: fmt, clippy -D warnings, check_rust_arch.py, orbyn-core store and crew tests, orbynd route tests over real HTTP, generated bindings committed.
• Not tested: none

Co-authors

none

AI usage

• Written by hand, with no AI
• Harness and version: Claude Code 2.1.321
• Models: claude-sonnet-5-5 (all)
• Effort: medium
• Human role: assigned the task and its boundaries; review pending
• Sources: none

Checklist

• Works on Windows and macOS
• web/src/generated with any contract change; docs/architecture.md with any rule change
• No copied source, secrets, credentials or personal data; no AI attribution in commits, title or branch