{"version":"2.1.288","anchor":"tool-result-and-error-text-scanned-for-secrets-in-host","canonical_anchor":"tool-result-and-error-text-scanned-for-secrets-in-host","heading":"Failed tool calls get an extra handling step, and tool calls record their permission","tier":"internal","area":"Tools","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.288\/e\/tool-result-and-error-text-scanned-for-secrets-in-host","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.288","markdown":"### Failed tool calls get an extra handling step, and tool calls record their permission\n\nWhen a tool call fails, Claude Code can pass its input and error text to a new handling step, and tool calls now record their permission\n\n**Unclear.** What the new handling step does with the error text, and what condition switches it on, are not established.\n\n**What**\n\nA tool call is one action Claude takes, such as running a command or editing a file. When a tool call ends in an error and a certain internal condition is met, Claude Code now passes the tool's input and the error text to a new handling step. Separately, each tool call now records who allowed it and which permission was checked for it. The usage data about permission decisions is now worked out from those records.\n\n**Why**\n\nThe way Claude Code tracks permissions for each tool call has changed underneath. Nothing here describes a new prompt or a new choice for you.\n\n- Area: Tools\n- Tier: Under the hood\n- Useful: 1\/5\n- Signal: 2\/5\n- Scope: individual\n- Heads-up: no"}