WebChat TTS reports success but produces no audible playback or downloadable audio (Telegram works)

Problem

In WebChat, TTS tool calls return success ("Generated audio reply"), but no audio is heard and no media/file is surfaced to the user.

Environment

• OpenClaw 2026.3.28 (updated from 2026.3.13)
• macOS host (Core), WebChat Control UI (openclaw-control-ui)
• messages.tts.provider = "openai"
• Telegram channel enabled and healthy

Repro

1. Open WebChat session.
2. Trigger TTS for a short phrase (e.g. “Don’t worry. It’s all fine!”).
3. Observe tool success message.
4. Run openclaw logs --follow during the trigger.

Expected

WebChat should provide audible playback and/or downloadable audio media, with clear routing logs.

Actual

• No audible playback in WebChat
• No downloadable media/file path
• No obvious TTS playback event in gateway logs for that trigger

Control test

Same TTS works on Telegram (audible result confirmed). Logs show Telegram send success (gateway/channels/telegram ... sendMessage ok ...).

Impact

TTS appears channel-specific: usable on Telegram, unreliable/opaque on WebChat.

Log snippet (exact excerpt from our run)

15:13:58+00:00 info agents/model-providers {"subsystem":"agents/model-providers"} [xai-auth] bootstrap config fallback: no config-backed key found
15:15:04+00:00 info agents/model-providers {"subsystem":"agents/model-providers"} [xai-auth] bootstrap config fallback: no config-backed key found
15:15:37+00:00 info agents/model-providers {"subsystem":"agents/model-providers"} [xai-auth] bootstrap config fallback: no config-backed key found
15:15:41+00:00 info gateway/channels/telegram {"subsystem":"gateway/channels/telegram"} telegram sendMessage ok chat=609356996 message=276
15:16:57+00:00 info agents/model-providers {"subsystem":"agents/model-providers"} [xai-auth] bootstrap config fallback: no config-backed key found