Telegram Multilingual Voice Reply
PassAudited by ClawScan on May 1, 2026.
Overview
This is a coherent Telegram voice-reply skill that uses local ASR/TTS scripts and Telegram messaging as described, with some normal but important dependency, messaging, and temporary-file considerations.
This skill appears safe for its stated Telegram voice-reply purpose. Before installing, make sure you are comfortable with it sending Telegram replies on your behalf, running local ASR/TTS helper commands, installing or using mlx_audio/ffmpeg/model dependencies, and leaving temporary audio or transcript files on your device.
Findings (4)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
The agent can post replies in Telegram on your behalf when this workflow is used.
The skill sends Telegram messages using the user's OpenClaw/Telegram messaging capability. This is expected for a Telegram reply workflow, but it is delegated account action.
用 OpenClaw `message` 工具发送（voice note + caption）：- `asVoice: true` - `path/filePath`: 上一步生成的 `reply.ogg` - `caption`: **同一段 reply_text**
Install only if you want OpenClaw to send Telegram replies for you, and use the provided text-only or voice-only overrides when needed.
You may need to install and trust local ML tooling and model downloads before the skill works.
The workflow relies on external ML packages/models, while the registry lists no install spec or required binaries. This is purpose-aligned, but dependency provenance and installation need user review.
STT 默认：`mlx-community/Qwen3-ASR-0.6B-8bit` ... `failed to import mlx_audio`：当前 `python3` 环境里没有安装到 `mlx_audio`。
Use a trusted Python environment, verify the mlx_audio package and model sources, and document/install ffmpeg and model dependencies explicitly.
The skill will run local Python/CLI tools to process audio when voice handling is requested.
The helper runs local CLI commands for ASR processing. This command execution is disclosed and central to the voice transcription purpose.
cli = "mlx_audio.stt.generate" ... p = subprocess.run(cmd, capture_output=True, text=True)
Run it in a normal user environment, keep dependencies updated, and avoid installing untrusted replacement binaries named mlx_audio or ffmpeg.
Sensitive voice transcripts may remain temporarily or persistently on the local machine in temp folders.
In CLI fallback mode, transcription output is written to a local temporary directory before being read back, and the artifact does not show cleanup of that directory.
out_prefix = Path(tempfile.mkdtemp(prefix="mlx_asr_out_")) / "out" ... txt_path = Path(str(out_prefix) + ".txt")
Use the skill on a trusted device and periodically clear temporary files if voice messages may contain sensitive information.