{"version":"2.1.317","anchor":"auto-mode-classifier-too-long-denials-marked-selfmending","canonical_anchor":"auto-mode-classifier-too-long-denials-marked-selfmending","heading":"Auto mode marks a too-long-conversation denial as one Claude Code will try to fix","tier":"notice","area":"Auto Mode","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.317\/e\/auto-mode-classifier-too-long-denials-marked-selfmending","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.317","markdown":"### Auto mode marks a too-long-conversation denial as one Claude Code will try to fix\n\nWhen auto mode can't review an action because the conversation is too long, Claude Code marks that denial as self-mending\n\n**What**\n\nIn auto mode, a classifier (a separate check run by a model) reviews each action before Claude Code carries it out. Sometimes the conversation is too long for the classifier to review an action, and the action is denied. That denial is now marked as self-mending. Its message says Claude Code will try to compact the conversation before the next request. Compacting means replacing earlier parts of the conversation with a shorter summary.\n\nA second kind of denial, given when the classifier's request fails with an HTTP error, is also marked as self-mending, but only in some cases.\n\n**Why**\n\nThe marking tells Claude Code that this denial can be recovered from, for example by compacting and trying again, so a long conversation in auto mode can carry on after it.\n\n- Area: Auto Mode\n- Tier: You'll notice\n- Useful: 2\/5\n- Signal: 2\/5\n- Scope: individual\n- Heads-up: no"}