All pull requests

Pull requests

• fix: measure e2e latency for realtime agents

  review_effort:high
  #7719·chenghao-mou opened Oct 10, 2026Member·

  ·
  6

• feat(plugins): add dotwave STT plugin

  #7718·Paul-HenriBJT opened Oct 10, 2026·

  ·
  6

• fix(voice): end the turn on an STT final that arrives after END_OF_SPEECH was consumed

  #7717·kunalgoesbyken opened Oct 10, 2026·

  ·
  1

  4

• fix: clean up unstarted async tee peers

  #7716·liangzhechen655 opened Oct 10, 2026·

  ·
  5

• fix: preserve nullable literal response schemas

  #7715·liangzhechen655 opened Oct 10, 2026·

  ·

• [draft] feat(agents): add DecisionModel for background decisions

  #7714·tinalenguyen opened Oct 10, 2026Member·

  ·

• fix: discard resampler history when resetting audio buffer

  #7713·liangzhechen655 opened Oct 10, 2026·

  ·

• fix: await sibling toolset cleanup before propagating close errors

  #7712·liangzhechen655 opened Oct 10, 2026·

  ·
  1

• fix: prevent zero-sized AudioByteStream frame loops

  #7711·liangzhechen655 opened Oct 10, 2026·

  ·
  1

• feat(llm): report prompt-cache write tokens as cache_creation_tokens

  #7710·adrian-cowham opened Oct 10, 2026Contributor·

  ·
  

• feat: add Zoom AI STT plugin

  #7708·cdliao opened Oct 9, 2026·

  ·
  1

• fix(inference): send 24 kHz for OpenAI STT models

  #7706·shawnfeldman opened Oct 9, 2026Contributor·

  ·
  2

• fix(nabrah): survive a mid-call reconnect, and honour flush

  #7703·tinalenguyen opened Oct 9, 2026Member·

  ·
  5

• fix(voice): make commit_user_turn await turn processing

  #7702·lumina678 opened Oct 9, 2026·

  ·
  1

  12

• fix(elevenlabs): fail fast on 4xx errors in STT recognize()

  #7701·pradneshfernandez opened Oct 9, 2026·

  ·
  11

• fix(voice): answer realtime tool calls interrupted past the timeout

  #7700·Li-john1021 opened Oct 9, 2026·

  ·
  1

  7

• fix(soniox): send the API key in the websocket handshake header

  #7696·mihafabcic-soniox opened Oct 9, 2026Contributor·

  ·
  2

• fix(elevenlabs): log idle text-to-dialogue timeouts at debug

  #7695·tim-w1d opened Oct 9, 2026·

  ·
  5

• fix(voice): map provider timestamps on the pushed-audio timeline for stt turn metrics

  #7694·15054538509 opened Oct 9, 2026·

  ·
  1

  12

• fix: flush buffered tts audio before playback drains

  review_effort:high
  #7692·chenghao-mou opened Oct 9, 2026Member·

  ·
  1

• fix(voice): keep a duplex model's input clock running while input audio is disabled

  #7691·TANGBUDU opened Oct 9, 2026·

  ·
  1

  4

• fix(voice): send silence to GPT-Live while input audio is off

  #7690·longcw opened Oct 9, 2026Contributor·

  ·
  1

  4

• fix(google): report unknown timing for results without word offsets

  #7687·15054538509 opened Oct 9, 2026·

  ·
  1

• fix(voice): wait for a paused activity before delivering a tool reply

  #7685·longcw opened Oct 9, 2026Contributor·

  ·
  1

  3

• feat(sarvam): support keyterm prompting on saaras:v4

  #7683·15054538509 opened Oct 9, 2026·

  ·
  1

  14