{"version":"2.1.323","anchor":"unicode-scan-failure-now-reports-a-descriptive-error-and-ret","canonical_anchor":"unicode-scan-failure-now-reports-a-descriptive-error-and-ret","heading":"A failed check for broken characters is now logged and retried","tier":"internal","area":"API Requests","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.323\/e\/unicode-scan-failure-now-reports-a-descriptive-error-and-ret","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.323","markdown":"### A failed check for broken characters is now logged and retried\n\nIf the scan for broken Unicode in a request fails, Claude Code now logs a clear error and runs the scan again from scratch\n\n**What**\n\nBefore sending a request to the model, Claude Code scans it for broken Unicode, such as half of a two-part character with its other half missing. If that scan itself fails, Claude Code now logs an error saying what happened. It then runs the scan again without its record of tool descriptions already known to be clean.\n\n**Why**\n\nThis rare failure can now be traced, and the request still gets checked.\n\n- Area: API Requests\n- Tier: Under the hood\n- Useful: 1\/5\n- Signal: 0\/5\n- Scope: individual\n- Heads-up: no"}