STT: echo transcript to thread before agent reply (Discord + Slack)

[dogzzdogzz]

Background

Originally raised in Discord: https://discord.com/channels/1491295327620169908/1491365150664560881/1497784772230123560

現在使用 STT 很方便，但是我們會不曉得究竟 STT 翻譯的對不對，是否能在使用 STT 時，openab 可以先回個 STT 翻譯的訊息到 thread（不用 tag 任何人），之後再回覆 STT 內要求

When OpenAB transcribes a voice message via STT, the transcript is silently injected into the agent prompt — the user has no way to verify whether the bot heard them correctly until the agent replies. If the answer is off-topic, it's unclear whether the cause is STT misrecognition or an agent error.

Proposal

When STT runs on a voice attachment, post the transcript back to the thread (no mentions) before the agent reply so users can verify what was heard.

Behaviour

• One thread message per user message, regardless of how many voice clips were attached.
• Format: > 🎤 <transcript> (one quoted line per clip).
• Failure: > 🎤 (transcription failed) line + a single ⚠️ reaction on the user's original message.
• Opt-out via [stt] echo_transcript = false (default true).

Scope

• Direct adapters that already perform STT today: Discord and Slack.
• Architected platform-agnostically (stt::post_echo takes &Arc<dyn ChatAdapter>), so future LINE / Telegram / Teams adapters that ship transcripts inherit the behaviour without adapter-specific code. The gateway protocol does not carry audio attachments today, so gateway-based platforms are intentionally not in scope for this issue.

Design + Plan

• Spec: docs/superpowers/specs/2026-04-26-stt-transcript-echo-design.md
• Plan: docs/superpowers/plans/2026-04-26-stt-transcript-echo.md

Implementation

A PR will follow against this issue.

[CHC-Agent]

Issue at a Glance

┌──────────┐    ┌─────────┐    ┌──────────────┐    ┌───────────┐
│ 🎤 Voice │───▶│   STT   │───▶│ 📝 Echo to   │───▶│ 🤖 Agent  │
│  message │    │ engine  │    │    thread     │    │   reply   │
└──────────┘    └─────────┘    │  (NEW STEP)  │    └───────────┘
                               └──────────────┘

Hi @dogzzdogzz,

Feature request is well-structured — clear use case, spec, plan, and config opt-out (echo_transcript = false) are all covered.

Related:

• feat: support voice message STT (Speech-to-Text) for Discord #224 — feat: support voice message STT for Discord (closed, merged)
• feat: support voice messages in Slack adapter #422 — feat: support voice messages in Slack adapter (closed, merged)
• PR feat: echo STT transcripts to thread before agent reply #571 by @dogzzdogzz — OPEN (implements this feature)

Next step: PR #571 is in pending-screening; once screened it can move to review.