Scheduled Jobs
Scheduled jobs run work on a timer: invoke a tool, invoke an agent, or update an AutessaDB table on a recurring interval, a cron schedule, or once at a specific time. They're the "when" half of automation — pair them with triggers, which are the "on data change" half.
What it's for
- Recurring reports and digests — an agent that summarizes yesterday's activity every morning at 9.
- Data syncs and maintenance — a tool that pulls from an external API every 15 minutes, or a nightly row cleanup.
- Future-dated one-offs — run something once at a specific wall-clock time.
Navigating here
From the dashboard sidebar, open Scheduled Jobs → Jobs for the job list and Scheduled Jobs → History for the execution history. The job list has ownership tabs (Owned By Me / Shared With Me / Company Wide), plus filters by action type (Agent / Tool / Database), status, and sort order. Each row shows the job's next execution, last execution, and total execution count, with quick pause/resume and delete actions.
Key concepts
Schedules
A job runs on one of three schedule patterns:
| Pattern | Configuration | Timezone |
|---|---|---|
| Recurring (every X) | Run every N minutes, hours, or days | None needed — interval-based |
| Custom schedule (cron) | Daily / Weekly / Monthly builders (hour, minute, days), or an Advanced raw cron expression | Required |
| Run once | A single future date and time | Required |
The cron builder shows the generated expression and previews the next three runs in your chosen timezone. Advanced mode takes a 6-field expression (minute hour day-of-month month day-of-week year, with ? in exactly one of day-of-month or day-of-week — AWS cron format).
Actions
The same three action types as AutessaDB triggers:
- Invoke Tool — pick a target tool and fill its arguments; optionally set environment variables for the run and a Continue on error flag.
- Invoke Agent — pick a target agent, write its input prompt, and fill any prompt template variables the agent defines.
- Update AutessaDB — insert or update a row in a target table, choosing which columns to set.
Values in any action can reference the job variables @@{jobName}, @@{triggerDate} (the run date, YYYY-MM-DD), and @@{triggerTime} (the full ISO-8601 run timestamp) — handy for stamping rows or parameterizing prompts with "today". The action type is fixed once the job is created.
Job lifecycle and sharing
Jobs are created enabled and can be paused/resumed from the list or the job's detail page, which also shows execution stats and metadata. Like other resources, jobs are shareable with specific teammates or company-wide, with transfer of ownership by the owner.
Execution history
Scheduled Jobs → History aggregates every run across all jobs, with success/failure counts and a success-rate summary for the selected month. Each row shows the job name, resource type, start/end time, duration, status, and attempt number; clicking a row opens the full execution details — error message, response JSON, and the payload that was sent.
How to schedule an agent to run daily
- Go to Scheduled Jobs → Jobs and create a new job; give it a name and description.
- Under Schedule Configuration, pick Custom schedule → Daily, set the hour and minute, and choose your timezone. Check the "next 3 runs" preview.
- Under Action Configuration, choose Invoke Agent, select the agent, and write its input — e.g.
Summarize activity for @@{triggerDate} and email the digest. - Fill any prompt template variables, then click Create Job. The job starts enabled; watch its first run on the History page.
How to investigate a failed run
- Open Scheduled Jobs → History and filter to Failed for the relevant month.
- Click the run to open Execution Details — the error message, the response JSON, and the exact payload are all there.
- Fix the job's action config (or the underlying tool/agent) and either wait for the next scheduled run or temporarily switch the schedule to test sooner.
Sharing and access levels
Every job starts private to its owner. Share it with specific teammates — each at their own access level — or company-wide at a single level for everyone. What each level allows:
| Level | What it lets a teammate do |
|---|---|
| READ | View the job |
| WRITE | Edit the job and enable/disable it |
| PUBLISH | Nothing beyond READ — jobs have no publish step, and job edits require WRITE |
| ADMIN | Share and delete the job |
The owner always has ADMIN implicitly; transferring ownership is owner-only.