Show HN: Ghost Vessel – give your local LLM agent a face (github.com)

A new project called "Ghost Vessel" has been announced, allowing users to integrate their local large language model (LLM) agents into video-call-style avatars. This innovative system creates a visually interactive avatar that serves as a user interface for personal AI agents like Hermes and OpenClaw, replacing traditional chat setups with a more engaging visual presence. The avatars are designed to display fine-grained facial emotions and persistent moods that adapt based on user interactions, enriching the communication experience with human-like expressions while keeping GPU usage minimal during runtime. Significantly, Ghost Vessel introduces a bridge that connects user commands directly to the agent, providing a seamless control surface that showcases dynamic command menus akin to those found in messaging platforms. This project emphasizes agent-agnostic compatibility, enabling users to create or incorporate various avatars—ranging from photorealistic to 2D and abstract styles—using a straightforward setup process. The technical implications include the ability for developers to create emotional and interactive AI agents that can maintain persistent states across sessions, enhancing user engagement and experience in AI interactions, all while being constructed with freely available tools and resources.