{"version":"2.1.311","anchor":"graceful-cake-quiet-turn-classifier","canonical_anchor":"graceful-cake-quiet-turn-classifier","heading":"Claude Code can now tell when a turn did only quiet background work","tier":"internal","area":"Background Agents","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.311\/e\/graceful-cake-quiet-turn-classifier","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.311","markdown":"### Claude Code can now tell when a turn did only quiet background work\n\nNew checks look at whether a turn ended on a status update or a shown decision, or did only scheduled or triggered work\n\n**Unclear.** Nothing shows what Claude Code does with the result of these checks.\n\n**What**\n\nClaude Code now has checks that look at how a turn ended. A turn is one round of Claude responding and acting. The checks ask:\n\n- whether the last tool call (an action Claude takes, such as running a command) was a decision shown to you, a status update, or something similar\n\n- whether the turn only did scheduled work or work started by a trigger\n\nThey depend on a server-controlled flag and on a particular tool from a connected MCP server being present. MCP servers are outside programs that give Claude extra tools.\n\n**Why**\n\nThis looks like groundwork for sorting quiet turns from ones that need attention in sessions that run unattended.\n\n- Flag `tengu_graceful_cake`: Not enough to say (read for one account on one subscription tier against v2.1.311; this account: no value returned, anonymous baseline: no value returned, compiled default: on) These values were read against a different version of Claude Code, so treat them as the nearest reading available instead of one taken on this release.\n- Area: Background Agents\n- Tier: Under the hood\n- Useful: 1\/5\n- Signal: 3\/5\n- Scope: individual\n- Heads-up: no"}