a-mart/cron-scheduling

Cron Scheduling

Use this skill when the user asks to schedule, reschedule, or cancel reminders/tasks for later.

Before creating a schedule, confirm:

• exact schedule timing (cron expression),
• timezone (IANA, for example America/Los_Angeles),
• task message content.

If the request is ambiguous, ask a follow-up question before adding a schedule.

Storage

• Schedule data is stored under ${SWARM_DATA_DIR}.
• Do not hardcode schedule file paths; the scheduler CLI resolves the correct storage location for the selected manager.
• --session is required when adding a schedule. Every schedule must target a specific session.

Session ID guidance

Every new schedule must include --session.

• Workers should usually schedule against their owning manager session ID. This is available from the SWARM_SESSION_ID env var when present, or from the agent identity / session context already provided to the runtime.
• Managers should use their own session / agent ID.

Commands

Run the scheduler CLI from this skill directory:

node ./schedule.js add \
  --manager "manager" \
  --session "manager--s2" \
  --name "Daily standup reminder" \
  --cron "0 9 * * 1-5" \
  --message "Remind me about the daily standup" \
  --timezone "America/Los_Angeles"

One-shot schedule (fires once at the next matching cron time):

node ./schedule.js add \
  --manager "manager" \
  --session "manager--s2" \
  --name "One-time deployment check" \
  --cron "30 14 * * *" \
  --message "Check deployment status" \
  --timezone "America/Los_Angeles" \
  --one-shot

Remove a schedule:

node ./schedule.js remove \
  --manager "manager" \
  --id "<schedule-id>"

List schedules:

node ./schedule.js list --manager "manager"

--session is required for add.

--manager is optional. If omitted, the CLI will auto-select a manager when there is only one manager or when a known default manager is detected.

Output

All commands return JSON:

• Success: { "ok": true, ... }
• Failure: { "ok": false, "error": "..." }