xeon_asr

Automatically converts received voice messages to text via an external ASR service, supporting multiple audio formats and integrating with OpenClaw.

Do not install this blindly. Key things to consider before proceeding: (1) The package is mislabeled — it claims to be an ASR skill but installs a persistent TTS service that modifies ~/.openclaw/openclaw.json and enables systemd services. If you expected ASR/transcription, do not install. (2) The installer sets HF_ENDPOINT=https://hf-mirror.com (an undocumented external mirror) — if you must install, inspect or remove that setting and prefer official Hugging Face endpoints. (3) Run install.sh with --skip-start and review generated config.json and tts_config.json before starting services. (4) Inspect xdp-tts-service package source (pip target) and any model repos before allowing automatic model downloads. (5) Consider running the installer in a sandbox/container or a throwaway VM first, or only install if you explicitly want a local TTS service that will run continuously. (6) If you do install, keep backups of your OpenClaw config (install script does a backup but verify) and review the created systemd user unit files before enabling them.

Type: OpenClaw Skill Name: xeonasr Version: 0.1.4 The skill bundle implements a local TTS and voice cloning service but exhibits high-risk behaviors during its installation and configuration phases. Specifically, `setup_env.sh` attempts to install system-level dependencies using `sudo apt-get/yum`, and `configure_openclaw_integration.sh` programmatically modifies the core `openclaw.json` configuration file. While these actions are plausibly necessary for the stated purpose of setting up a local OpenVINO inference environment, the requirement for elevated privileges and the automated tampering with global configuration files represent a significant attack surface. No clear evidence of intentional malice, such as data exfiltration or backdoors, was found.