Tools in Scheduled Runs

Last updated 2026-11-14

How the tool set differs when a run is started by the scheduler.
scheduler-skill-creator is the high-level skill that creates a scheduler skill body and calls schedules(mode=write) to bind it. New recurring / one-shot requests should activate that skill, not call the low-level tool directly.
The daemon-side runtime watches ~/.config/agenvoy/{tasks,crons}.json with fsnotify and hot-reloads on Write / Create / Rename. Past-due tasks are auto-fired and removed on startup or reload; fire executes via runtime.SetRunner(app.RunSkill) → exec/subagent.Exec, an in-process subagent over the scheduler skill body.
A fire goes through exec/subagent.Exec in its top-level mode, so a scheduled run does not get the narrower subagent tool set: only edit_tool, test_tool, edit_skill, schedules and store_secret are excluded, no subagent charter is added, and the TUI-only Skills are unavailable. No parent run supplies a permission mode, so the run executes as allow-all and, as a subagent, is not limited to the working directory; a run on a cli- session has no client to answer a confirmation, so a call that would ask simply runs, including an irreversible command, while a call that needs an approval it cannot get fails: a sensitive path or a path outside $HOME is refused as not approved, and argv ['sudo', ...] or sandbox: false is refused for lack of password verification. A run on a tg- or dc- session asks in that chat. pkg_manage was Removed in v1.3.2; use run_command instead.
The TUI /cron and /task commands were removed; use /schedule, which lists cron and one-shot entries together (enter fires one now, d deletes it, add / edit by asking the agent), plus /sched-<name> to manually trigger an existing scheduler skill body.