Agent Scheduling
Claudette can persist agent wakeups and recurring routines in its app database. Scheduled prompts are owned by a chat session and fire through the same send pipeline as a normal user message, so tray state, notifications, transcript persistence, and provider selection stay consistent.
Agent Tools
Section titled “Agent Tools”Agents can schedule their own wakeups and routines through native tools:
| Tool | Purpose |
|---|---|
ScheduleWakeup | Schedule a one-shot wakeup with delaySeconds or fireAt, plus a prompt and optional reason. |
CronCreate | Create a routine from a standard 5-field cron expression in local time. |
CronList | List persisted wakeups and routines. |
CronDelete | Delete a routine by id or name. |
Monitor | Subscribe to future output lines from a background task (Bash run_in_background, a backgrounded Agent/subagent, or a Workflow) instead of polling. |
Claude and Codex agents both reach these tools over the built-in Claudette MCP bridge. Scheduling stays available regardless of the Agent Attachments plugin toggle or the experimental Claudette MCP interaction tools.
Claude Code also ships its own built-in ScheduleWakeup / Cron* tools, but those keep schedules inside the CLI process — they don’t survive a restart and Claudette can’t see them. Whenever the Claudette MCP bridge is attached, Claudette hides those built-ins (--disallowedTools) so every schedule an agent makes lands in Claudette’s store and shows up in the Loops and Schedules view.
Wakeups and routines survive app restarts. If a fire time passes while Claudette is closed, the scheduler fires the overdue task when the app starts again.
Background task completion
Section titled “Background task completion”When an agent starts work in the background — a Bash command with run_in_background, a backgrounded Agent/subagent, or a Workflow — Claudette watches the task and automatically re-invokes the agent when it finishes, so the agent reports the result back to you without needing a manual nudge. This works for every background task type the agent can launch, not just Bash, and long-running tasks are waited on until they complete rather than being abandoned.
If the owning workspace is archived, deleted, or otherwise has no worktree when a task comes due, Claudette pauses the task instead of retrying it forever. The Automation settings row and routine list payload include the disabled reason and last error so you can delete or recreate the routine deliberately.
The CLI exposes routines for automation and scripts:
claudette routine listclaudette routine create <session-id> "0 9 * * 1-5" "Check open PRs" --name weekday-prsclaudette routine create <session-id> "30 14 28 2 *" "Review the launch checklist" --onceclaudette routine run weekday-prsclaudette routine pause weekday-prsclaudette routine resume weekday-prsclaudette routine delete weekday-prsroutine create accepts literal prompts, @file prompts, and - for stdin, matching chat send. routine resume also re-arms a task that was switched off automatically (archived workspace, dispatch error) and clears its failure count. A one-shot wakeup that already fired can’t be resumed — use routine run to fire it again.
Loops and Schedules view
Section titled “Loops and Schedules view”You can drive the same scheduling backend from the GUI without the agent involved.
Click the clock icon next to the dashboard icon in the sidebar header to open the Loops and Schedules view. Like the Dashboard, it keeps the sidebar and header (panel toggles) in place — click the clock again, the dashboard icon, or a workspace to leave it.
The view is a live monitor for everything scheduled across every workspace and backend (Claude and Codex), whether you created it, an agent did, or the CLI did. It updates on its own as tasks are created, fire, fail, or change — no refresh needed.
Summary strip. Counts of armed loops, armed one-shots, runs in progress right now, and tasks that need attention, plus which task fires next and when. Click the attention tile to filter down to just those.
Filters. Narrow by status (All, Active, Paused, Needs attention), by workspace, or by searching names, prompts, and schedules. Clear finished deletes one-shots that already fired.
The list itself is split into:
- Loops — recurring cron routines.
- Schedules — one-shot wakeups and run-once cron routines.
Each row shows:
- A status: Running (its session has a turn in progress), Active, Failing ×N (still armed, but recent fires errored), Off (switched off automatically, with the reason — e.g. the workspace was archived), Paused, or Done (a one-shot that already fired).
- The schedule in plain language next to the raw cron expression, a countdown bar to the next fire, and when it last ran.
- The backend and model the fired turn will run on, when the task pins one.
- The target workspace › session. Click the session to jump to it. For new session each run tasks, open next to “last run” jumps to the session the most recent fire created.
- The last error inline when a task is failing or off.
Expand a row to see the full prompt, the reason (for agent-created wakeups), and its recent runs — each with its outcome, whether it was triggered manually, and a link to the session it landed in.
Row actions: Run now (same as routine run), Pause / Resume, and Delete (click twice to confirm). Resuming re-arms the task at its next cron match (or its original fire time, firing immediately if that passed while paused) and clears the failure count.
At the bottom, Recent activity lists the latest fires across every task.
The clock icon in the sidebar carries a badge with the number of armed tasks. It turns into a ! when something needs attention, and pulses while a scheduled run is in progress. The same list (without the monitoring extras) also appears under Settings → Automation.
Creating a task from the GUI
Section titled “Creating a task from the GUI”The New scheduled task button at the top of the page opens a dialog with two modes:
- One-shot (wakeup) — pick a target, a date/time, and a prompt.
- Recurring (cron) — pick a target, write a standard 5-field cron expression (e.g.
0 9 * * 1-5), optionally name the routine, choose recurring or one-shot, and write the prompt.
Choosing a target. The dialog has a cascading picker: select a repository, then a workspace in it, then either an existing session in that workspace or ✦ New session each run. New-session tasks create a brand-new chat session in the workspace every time they fire — so a recurring loop gets a clean session per run instead of piling onto one. (The cascade can target any workspace, not just the one you currently have open.)
Seeing what fired
Section titled “Seeing what fired”When a task fires, the triggering prompt renders in the target session as a normal user message, tagged with a Scheduled badge so it’s clear the agent’s work was kicked off by the scheduler and not typed by you. The badge persists across reloads and survives session forking.
Slash commands
Section titled “Slash commands”In a chat, you can create tasks for the current session without opening the dialog:
/loop 5m run the test suite/schedule 2026-06-01T09:00 cut the release branch/schedule # opens the dialog (no args)/schedule write the weekly status update # opens the dialog, prompt prefilled/loop <interval> <prompt> translates an interval into a cron expression and creates a recurring routine in the current session. The interval must evenly divide its unit: minutes that divide an hour (e.g. 1m, 5m, 15m, 30m), hours that divide a day (e.g. 1h, 2h, 6h, 12h), or 1d. Any other interval (sub-minute, or one that doesn’t divide evenly) opens the dialog in cron mode so you can write the expression directly.
/schedule <when> <prompt> accepts a leading timestamp — YYYY-MM-DD, YYYY-MM-DDTHH:MM, or a date and time as two tokens (YYYY-MM-DD HH:MM) — interpreted in your local time. If both the time and prompt are present, it creates the wakeup inline; otherwise it opens the dialog prefilled with whatever you provided.