{"version":"2.1.314","anchor":"auto-mode-denials-tool-input-serialization-and-permission-d","canonical_anchor":"auto-mode-denials-tool-input-serialization-and-permission-d","heading":"Auto mode records blocked actions without minor details, and its denial message is rebuilt","tier":"notice","area":"Auto Mode","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.314\/e\/auto-mode-denials-tool-input-serialization-and-permission-d","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.314","markdown":"### Auto mode records blocked actions without minor details, and its denial message is rebuilt\n\nWhen auto mode blocks an action, Claude Code now records it without its description, timeout or switched-off options, and rebuilds the denial message\n\n**What**\n\nAuto mode is a way of working where Claude Code decides for itself whether an action is safe to allow. When it blocks an action, Claude Code now makes a record of what was asked. The record leaves out:\n\n- the action's description\n\n- its timeout\n\n- sandbox or background options that are switched off\n\nIf that record cannot be made, Claude Code logs an error saying so. The message Claude gets when auto mode refuses permission has also been rebuilt, including the part about the auto mode consent step.\n\n**Why**\n\nTwo blocked requests that differ only in those details now produce the same record, which likely lets Claude Code treat them as the same denial.\n\n- Area: Auto Mode\n- Tier: You'll notice\n- Useful: 1\/5\n- Signal: 1\/5\n- Scope: individual\n- Heads-up: no"}