This repository was archived by the owner on Jul 16, 2026. It is now read-only.
Repository navigation
feat(voice-agent): port [BLANK_AUDIO] filler token (Phase 5.4) - #74
Merged
Merged
Conversation
Port the only clean OSS sutando win from the voice-agent.ts reconciliation (Phase 5.4): add the "[BLANK_AUDIO]" token to the FILLERS-ARE-NOT-REQUESTS backchannel list so a transport-emitted blank-audio marker is ignored rather than (mis)interpreted as a directive that triggers `work`. The rest of OSS's voice-agent.ts diff was reviewed and intentionally NOT ported: stando is ahead on multi-provider voice/STT + the cloud gateway, uses state-paths.js (not workspace_default.js) by design, and the read_selection deictic prompt depends on a tool stando doesn't have. Co-authored-by: Cursor <cursoragent@cursor.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Phase 5.4 of the OSS→private
voice-agent.tsreconciliation. After a full diff review, the only clean, safe OSS win was a single prompt token: add"[BLANK_AUDIO]"to the FILLERS-ARE-NOT-REQUESTS backchannel list so a transport-emitted blank-audio marker is ignored instead of being (mis)interpreted as a directive that fireswork.What was reviewed and intentionally NOT ported
voice-agent.tshas diverged at the architectural foundation; in most areas stando is ahead:VOICE_PROVIDER/OpenAI Realtime),SUBAGENT_PROVIDER,STT_PROVIDER+ Whisper, the cloud gateway (managed-Gemini minting,cloudRecordEvent/skill.runmetering, session lifecycle),state-paths.jsresolution.recordToolCall([conversation-store] remove triple-encoded redundancy after per-surface tables (#1051 follow-up) sonichi/sutando#1052) — stando replaced local-sqlite tool recording with richer cloud telemetry;workspace_default.jsresolver — stando usesstate-paths.jsby design;voice-agent.config.jsontemplating — stando uses env vars by design.read_selection/ DEICTIC-screen-reference prompt depends on aread_selectiontool +setSessionToolUpdaterthat don't exist anywhere in stando.Test plan
pnpm run typecheckpasses[BLANK_AUDIO]utterance no longer triggers aworkdispatch in a live voice sessionMade with Cursor