Repository navigation
feat: wait for voice-note transcriptions before agent turns - #83
Draft
dimavrem22 wants to merge 6 commits into
Draft
dimavrem22 wants to merge 6 commits into
dimavrem22 wants to merge 6 commits into
Conversation
dimavrem22
force-pushed
the
feat/voice-note-transcriptions
branch
from
November 18, 2026 03:00
c219b98 to
026febe
Compare
dimavrem22
force-pushed
the
feat/voice-note-transcriptions
branch
from
November 18, 2026 08:03
026febe to
02d2afb
Compare
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 join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Executive Summary
Wait for voice-note transcription before waking Codex, while retaining the original attachment fallback.
0.2.36.Description
One subscription to
message.transcription_finishedcovers MMS, iMessage, and WhatsApp. The receiver matchesdata.channelanddata.message_idto the original waiting message; a completion does not create another turn.Managed deliveries use the already-resolved voice-note result without a second wait. Standalone delivery retains its 60-second completion wait and single bounded read. Signed attachment transcript metadata survives URL refresh.
Pending audio receipts are persisted and acknowledged immediately. A matching completion notification updates the same original message rather than creating another turn. A background consumer resumes the existing channel handler when every transcription is terminal, or after the original 60-second deadline plus one bounded read. Ordered conversation consumers use the same resolution path after their existing durable admission.
Generated speech is explicitly labeled as a voice-note transcript, separate from typed text. Original attachments, sender text, reply destinations, and existing session admission remain intact; unavailable transcripts continue through the previous attachment-handling path.
Reason
Agents should understand voice notes without racing their transcription or answering twice when completion arrives later.
Decisions
origin/main; no deployments are included.Testing
Fable and Opus reviewed the prior transcription implementation; those reviews predate the unified webhook change.
Unified webhook head:
01c8e3680. GitHub checks for this PR. No live-environment verification or rollout.Local verification is static only: applicable lint/format, syntax and diff checks. No local tests or CI-equivalent builds were run.
Regression coverage includes pending completion, timeout fallback, restarts, source-bound metadata and no second hosted wait where applicable.
No deployments, live-message sends, shared-database mutations or package publication are included.
Related PRs
feat: deliver voice-note transcripts before agent turns deepseek-harness-plugin#10
Discover remote tools and preserve startup and reply identity #82
feat: expose audio attachment transcription results inkbox#247
Open in [code]smith · Need help on this PR? Tag
@codesmith-botwith what you need.