All issues

google(realtime): on models without the reply placeholder, generateReply instructions can be recited as model speech

#2589

· pivvix opened

on Sep 25, 2026

google(realtime): commitAudio()/clearAudio() are no-ops, so manual activity detection cannot close a user turn

#2581

· pivvix opened

on Sep 24, 2026

voice: TranscriptionSynchronizer rotated-out segments cause duplicates

#2566

· berk-h opened

on Sep 22, 2026

Expose an early latched JobContext cancellation signal so pending startup can stop when the job terminates

#2545

· CampbellMBXJ opened

on Sep 21, 2026

Job-process supervisor's ping/pong watchdog fires on a live, non-hung process — reproduced both mid-teardown and fully idle

#2533

· seniojuk opened

on Sep 18, 2026

Unhandled promise rejections from `.finally(...)` on task results, and a lint rule that would catch them

#2508

· Rehansanjay opened

on Sep 16, 2026

Question: expose raw Gemini Live UsageMetadata in the Node.js Google realtime plugin

#2490

· yagince opened

on Sep 14, 2026

deepgram/xai STT plugin: unexpected WebSocket close is never detected, reconnect never fires (bare Task in Promise.all)

#2469

· obaverstam opened

on Sep 10, 2026

Unplayed response resumes while nonempty preflight is still awaiting final transcript (1.8.0)

#2430

· sherifsarhan opened

on Sep 6, 2026

AudioRecognition startedSpeakingAt can inherit a late VAD end timestamp (1.8.0)

#2429

· sherifsarhan opened

on Sep 6, 2026

Feature: per-tool execute timeout, a hung blocking tool stalls the turn forever

#2428

· HoneyTyagii opened

on Sep 5, 2026

TranscriptionSynchronizer should flush text for zero-duration audio segments

#2383

· coderabbitai opened

on Aug 31, 2026