{"version":"2.1.318","anchor":"context-clearing-reports-cleared-tool-uses-and-measured-toke","canonical_anchor":"context-clearing-reports-cleared-tool-uses-and-measured-toke","heading":"Context clearing reports more detail about cleared tool results and token counts","tier":"internal","area":"Context Management","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.318\/e\/context-clearing-reports-cleared-tool-uses-and-measured-toke","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.318","markdown":"### Context clearing reports more detail about cleared tool results and token counts\n\nAfter old tool results are cleared, Claude Code now records how many tool uses went, the token count, and whether that count was measured\n\n**What**\n\nContext is the conversation Claude can see at once, measured in tokens (small pieces of text). When old tool results are cleared to free room, Claude Code now reads three things from the report instead of one number:\n\n- the input tokens involved\n\n- how many tool uses were cleared\n\n- whether the token count was actually measured\n\nPreviously it used a single number, the smaller of the server's and its own estimate. Now, when its own estimate was measured from the prompts, it trusts that estimate. The record kept when results are cleared after a period of inactivity now also notes the token count used and whether it was measured, and the condition that decides when that record is sent no longer looks at an out-of-date value.\n\n**Why**\n\nClaude Code's idea of how full the context is after tool results are cleared is more accurate.\n\n- Area: Context Management\n- Tier: Under the hood\n- Useful: 1\/5\n- Signal: 1\/5\n- Scope: individual\n- Heads-up: no"}