omni-media
LINJIANG12
Universal multimodal audio/video MCP servers for AI agents lacking built-in audio tools (zcode, opencode, dsh, …): host-native listening (read_audio) uses the host's native audio model; external delegation (read_media) calls an audio-model API. 为不支持音频工具的 Agent 提供两种音频读取 MCP,分别适配宿主原生音频模态与外部音频模型 API。