moss-tts-family-chatbot
AdvisoryAudited by Static analysis on Apr 30, 2026.
Overview
No suspicious patterns detected.
Findings (0)
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.
Audio attachments may be sent for transcription as soon as they appear in a conversation.
The skill tells the agent to automatically prioritize transcription when media markers appear. This is expected for an ASR skill, but it is a strong automatic trigger.
消息中包含 "[media attached:"、"audio/ogg"、"audio/opus" ... 时——用户发来了语音，必须立刻转写，禁止说"我没有语音识别能力"。
Keep this trigger limited to real user-provided media attachments, and ask for confirmation if the attachment origin or user intent is unclear.
The skill could help create unauthorized voice impersonations, which may mislead listeners or create legal and trust risks.
The instructions encourage obtaining an online sample of a named celebrity's voice and using the voice-clone API, without any consent, authorization, or impersonation safeguards.
如果用户需要生成特定的某个名人的声音，可以告诉用户自己需要上网找一段音色，然后用这个音色调用voice clone接口
Require explicit authorization for voice cloning, refuse requests to clone public figures or private individuals without consent, and document acceptable-use limits.
If configured, the skill has delegated authority to post audio messages into Feishu chats, which is a third-party account action beyond simple local audio generation.
The script uses Feishu app credentials to obtain a tenant token and then sends a chat message, while the registry metadata declares no required env vars or primary credential.
-d "{\"app_id\":\"${FEISHU_APP_ID}\",\"app_secret\":\"${FEISHU_APP_SECRET}\"}" ... "https://open.feishu.cn/open-apis/im/v1/messages?receive_id_type=chat_id"Declare all required credentials and scopes, validate that Feishu credentials are intentionally configured, use the least-privileged bot scopes, and require clear user confirmation for the target chat.
Voice recordings and their contents leave the local environment for processing by the MOSI service.
The ASR script uploads the selected audio file to the MOSI provider for transcription. This is purpose-aligned, but audio can contain sensitive personal or business content.
"https://studio.mosi.cn/api/v1/audio/transcriptions" ... -F "file=@${FILE}"Tell users when audio will be uploaded, avoid processing sensitive recordings unless authorized, and review MOSI retention/privacy terms.