{"version":"2.1.281","anchor":"transcript-resume-adopts-recorded-wire-tool-input-copies","canonical_anchor":"transcript-resume-adopts-recorded-wire-tool-input-copies","heading":"Transcript resume adopts recorded wire tool-input copies","tier":"internal","area":"Resume","url":"https:\/\/changelogs.core-directive.com\/v\/2.1.281\/e\/transcript-resume-adopts-recorded-wire-tool-input-copies","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.281","markdown":"### Transcript resume adopts recorded wire tool-input copies\n\nResuming a session now reuses the saved record of tool inputs sent to the API, but only where it can prove they still match\n\n**Unclear.** The finding does not say what a reader notices when a resumed session reuses these records.\n\n**What**\n\nWhen Claude uses a tool, its input is sent to Anthropic's API in a particular form. Claude Code now records that exact form in the session's transcript, the saved history of the conversation. When you resume a session, it reuses that record only for inputs it can prove still match. Anything it cannot prove is checked again while the session runs. The debug log shows lines such as \"wire tool input copies: adopted N from the transcript's record\".\n\n**Why**\n\nA resumed session only trusts saved tool inputs it has confirmed, and it checks the rest again instead of assuming they match.\n\n- Area: Resume\n- Tier: Under the hood\n- Useful: 1\/5\n- Signal: 2\/5"}