Bug type

Behavior bug (incorrect output/state without crash)

Summary

1. Bug Description

3. When using the Feishu channel, TTS audio is truncated when:

4. • User sends message via Feishu
5. • AI replies with TTS audio to Feishu
6. • Only a few words are played (e.g., 2-3 words)

8. However, when using WeCom to send message and Feishu to receive reply, the TTS audio plays correctly.

12. Environment

13. • OpenClaw version: 2026.2.26
14. • Channels: Feishu + WeCom
15. • Gateway mode: local

17. Additional Notes

18. • TTS files are generated correctly (file size ~20KB)
19. • Issue only occurs when Feishu is both source and target
20. • Cross-platform (WeCom -> Feishu) works correctly

Steps to reproduce

10. Steps to Reproduce

11. 1. Configure both Feishu and WeCom channels
12. 2. Set messages.tts.auto: "always"
13. 3. Send a message from Feishu
14. 4. Reply is sent as TTS audio
15. 5. Audio is truncated to only 2-3 words

Expected behavior

17. Expected Behavior

18. Full TTS audio should be played.

Actual behavior

3. When using the Feishu channel, TTS audio is truncated when:
4. • User sends message via Feishu
5. • AI replies with TTS audio to Feishu
6. • Only a few words are played (e.g., 2-3 words)

OpenClaw version

21. • OpenClaw version: 2026.2.26

Operating system

操作系统：OpenCloudOS 版本号：9.4 内核版本：6.6.117-45.1.oc9.x86_64 类似于 CentOS 9 / RHEL 9

Install method

No response

Logs, screenshots, and evidence

Impact and severity

No response

Additional information

No response