{"version":"2.1.299","anchor":"container-restart-recovery-offers-cut-off-agents-for-resume","canonical_anchor":"container-restart-recovery-offers-cut-off-agents-for-resume","heading":"After a container restart, Claude resumes stopped agents instead of starting new ones","tier":"notice","area":"Subagents","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.299\/e\/container-restart-recovery-offers-cut-off-agents-for-resume","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.299","markdown":"### After a container restart, Claude resumes stopped agents instead of starting new ones\n\nWhen a container restarts mid-turn, Claude is told which agents it launched were stopped and to resume them by messaging them\n\n**What**\n\nAgents are helper instances of Claude that it launches to work on part of a task. If the container running a session restarts in the middle of a turn, Claude now gets a notice. The notice says the container was restarted, lists the agents launched in that turn, and says they have stopped. It tells Claude to resume each one by sending a message to its agent id rather than launching it again.\n\n**Why**\n\nRemote sessions recover better from a restart, keeping the agents' earlier work instead of starting over.\n\n- Area: Subagents\n- Tier: You'll notice\n- Useful: 2\/5\n- Signal: 2\/5\n- Scope: individual\n- Heads-up: no"}