{"version":"2.1.284","anchor":"schedulewakeup-loop-tool-results-now-tell-the-model-to-pos","canonical_anchor":"schedulewakeup-loop-tool-results-now-tell-the-model-to-pos","heading":"Scheduled wakeups now tell Claude to post a status update before ending its turn","tier":"notice","area":"Slash Commands","url":"https:\/\/changelogs.core-directive.com\/v\/2.1.284\/e\/schedulewakeup-loop-tool-results-now-tell-the-model-to-pos","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.284","markdown":"### Scheduled wakeups now tell Claude to post a status update before ending its turn\n\nAfter scheduling a wakeup in `\/loop`, Claude is now told to post any owed status update and then end its turn\n\n**Unclear.** What decides when the \"only thing Claude does\" wording is used is not settled.\n\n**What**\n\nWhen Claude schedules its next wakeup with the ScheduleWakeup tool (used by `\/loop` to run work on a repeating timer), the text Claude gets back from the tool changed. It used to say there was nothing more to do this turn. It now tells Claude that if it owes you a status update this round, it should give it and then end its turn.\n\n- When scheduling the wakeup is the only thing Claude does in that step, it is told the turn ends as soon as the tool returns, so any update must come before the call.\n\n- In brief mode, updates must go through the message tool, because plain reply text counts as unread there.\n\n- The messages for a stopped loop and for a loop reaching its maximum duration now also tell Claude to end the turn.\n\nThe tool's own description is unchanged in this release; only the text it returns after a call is different.\n\n**Why**\n\nIf you use `\/loop`, Claude should post its per-round updates more reliably, where before they could go missing.\n\n- Area: Slash Commands\n- Names: `\/loop`\n- Tier: You'll notice\n- Useful: 2\/5\n- Signal: 1\/5"}