{"version":"2.1.304","anchor":"subagent-permission-prompts-suppressed-after-an-unanswered-o","canonical_anchor":"subagent-permission-prompts-suppressed-after-an-unanswered-o","heading":"Prompts are held back with a quiet_spell outcome after one goes unanswered","tier":"notice","area":"Subagents","scope":"individual","heads_up":true,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.304\/e\/subagent-permission-prompts-suppressed-after-an-unanswered-o","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.304","markdown":"### Prompts are held back with a quiet_spell outcome after one goes unanswered\n\nAfter a permission prompt times out unanswered, later subagent prompts in the same turn return quiet_spell instead of asking again\n\n**Unclear.** It is not shown what the subagent does when its request is held back, for example whether it treats it as a refusal.\n\n**What**\n\nA permission prompt is the question Claude Code asks before it does something that needs your approval. A subagent is a helper agent that the main Claude starts to work on part of a task. Before, every permission request from a subagent raised its own prompt, and a prompt that timed out only noted that it had gone unanswered.\n\nNow, when a prompt in the same turn has timed out with no answer, a later request from a subagent does not raise another prompt. It returns the outcome `quiet_spell` instead.\n\n- The quiet period lasts as long as the prompt timeout.\n\n- A timed-out prompt is recorded as unanswered (`promptsUp.unanswered`), and answering any prompt with allow or deny clears that record.\n\n- The behavior is controlled by `tengu_hushed_plover`. When the server supplies no value for it, the code defaults to turning the behavior on.\n\n- `quiet_spell` is also a new reason in web fetch provenance checks, the checks about where fetched web content came from, and it is recorded separately there.\n\n**Why**\n\nIf you step away while Claude is working, subagents no longer pile up repeated approval requests within one turn. A subagent action may simply not happen during that window, so an unanswered prompt can mean later subagent steps were skipped rather than queued for you.\n\n- Flag `tengu_hushed_plover`: Not enough to say (read for one account on one subscription tier against v2.1.304; 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: Subagents\n- Tier: You'll notice\n- Useful: 2\/5\n- Signal: 3\/5\n- Scope: individual\n- Heads-up: yes"}