Schedules & Automation
Run tenant-owned DeepSpace prompts on a controlled interval, keep durable run history, and pause or cancel work without losing auditability.
User-owned schedules
Each schedule stores a name, prompt, interval, active state, next-run time, owner, and tenant boundary.
Durable run history
Runs move through queued, running, completed, failed, or cancelled states and remain queryable through the schedule history API.
Worker-safe dispatch
Celery Beat checks due schedules and workers claim work with database locking so two workers do not execute the same run.
Control and audit
Pause, resume, update, delete, and cancellation actions remain authenticated, rate-limited, tenant-scoped, and auditable.
Lifecycle
active schedule → Beat dispatch → queued run → worker claim
→ DeepSpace execution → result/artifact → next-run updateHow to use it
- Open Schedules from the DeepSpace header.
- Enter a focused prompt and interval, then create the schedule.
- Pause it while editing, or cancel a run that is still in progress.
- Open run history to inspect the outcome and any generated artifacts.
Operational requirements
The API can be deployed without enabling schedules, but automatic dispatch requires the worker and Celery Beat services to be running against the same database and broker. Every deployment should set its own rate, retention, notification, and concurrency limits before exposing schedules to a large tenant population.