Claude· Updated
Claude Code now supports dynamic scheduling for recurring tasks, allowing the agent to adjust its own check-in intervals based on activity. By using the new Monitor tool, the agent can watch background processes directly instead of polling, making autonomous maintenance and CI triage more efficient.
/loop command without a fixed interval allows the agent to intelligently schedule its own ticks (the iterative cycles an agent follows) between one minute and one hour based on task urgency.This update shifts the agent from reactive polling to proactive monitoring. By using the new Monitor tool, Claude Code can stream output from background scripts in real-time rather than repeatedly re-running prompts. This reduces token consumption and latency when managing long-running builds, pull request reviews, or continuous integration failures.
Trigger a built-in maintenance mode by entering a bare /loop command to triage CI and address PR comments. For project-specific automation, define custom instructions in a .claude/loop.md file. These session-scoped tasks require version 2.1.72 or later and expire automatically seven days after creation.
Claude now supports dynamic looping. If you run /loop without passing an interval, Claude will dynamically schedule the next tick based on your task. It also may directly use the Monitor tool to bypass polling altogether /loop check CI on my PR https://t.co/L5UFH22r0C
Every HeadsUpAI update is written based on its original source and reviewed before it's published. Read our editorial standards →
