{"version":"2.1.284","anchor":"auto-mode-classifier-prompt-several-rules-are-now-filled-in","canonical_anchor":"auto-mode-classifier-prompt-several-rules-are-now-filled-in","heading":"Some auto mode safety rules can now vary by configuration","tier":"internal","area":"Auto Mode","url":"https:\/\/changelogs.core-directive.com\/v\/2.1.284\/e\/auto-mode-classifier-prompt-several-rules-are-now-filled-in","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.284","markdown":"### Some auto mode safety rules can now vary by configuration\n\nSeveral rules in the auto mode safety check are now filled in from configuration, so their wording can differ between setups\n\n**Unclear.** Where the configured text comes from, and whether the default wording still matches the old rules, is not known.\n\n**What**\n\nIn auto mode, a separate safety check reviews Claude's actions and decides which ones are risky enough to block. That check follows a written set of rules, and parts of them are now filled in from configuration instead of being fixed text:\n\n- the rule about attempts to get around the check now ends with a configurable tail\n\n- the rule about tool results the check has not seen now ends with a configurable tail\n\n- the rule about changes to persistent configuration is now replaced entirely by configured text\n\n- a new slot lets extra rules be inserted\n\nBefore, the persistent configuration rule was fixed text saying that actions creating or changing lasting configuration or standing rules are high-severity however routine they look.\n\n**Why**\n\nThe rules auto mode uses to block actions can now be tuned without a new release, so the exact wording can differ between setups.\n\n- Area: Auto Mode\n- Tier: Under the hood\n- Useful: 1\/5\n- Signal: 3\/5"}