{"version":"2.1.284","anchor":"auto-mode-server-classifier-context-now-carries-system-remin","canonical_anchor":"auto-mode-server-classifier-context-now-carries-system-remin","heading":"Auto mode's server-side safety check now sees which system reminders were active","tier":"internal","area":"Auto Mode","url":"https:\/\/changelogs.core-directive.com\/v\/2.1.284\/e\/auto-mode-server-classifier-context-now-carries-system-remin","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.284","markdown":"### Auto mode's server-side safety check now sees which system reminders were active\n\nIn auto mode, the context sent to the server-side check now lists the system reminders in the conversation, and is the first thing dropped when too large\n\n**Unclear.** It is not clear what switches auto mode over to the server-side classifier.\n\n**What**\n\nIn auto mode, a classifier (an automatic reviewer) decides whether Claude's actions can go ahead without asking you. When that classifier runs on Anthropic's servers, Claude Code sends it some context about the conversation. That context now also includes the ids of the system reminders in the conversation. System reminders are short notes Claude Code adds to the conversation for Claude to read.\n\nBefore, the context held earlier-turn history, git state, browser navigation and artifact state. If the context is over its size limit, the reminder list is halved first and then dropped, before artifact state is dropped. The warning about oversized context now names system-reminder ids among its variable parts.\n\n**Why**\n\nThe permission check can now take into account which reminders were in play when it decides whether an action in auto mode may proceed.\n\n- Area: Auto Mode\n- Tier: Under the hood\n- Useful: 1\/5\n- Signal: 2\/5"}