Multimodal MCP Client

Example Use Cases

• 1

  Accept and process voice, text, and image inputs to drive multimodal AI workflows.

• 2

  Orchestrate and chain tool-based operations via the Model Control Protocol (MCP) for complex, multi-step interactions.

• 3

  A developer-friendly, local or hosted setup to prototype voice-first AI applications with configurable MCP servers.

Description

An open-source Vite + TypeScript client that provides a modern voice-controlled interface for orchestrating AI workflows via the Model Control Protocol (MCP). It integrates Google Gemini for multimodal understanding and Systemprompt/Anthropic MCP tooling to execute extensible workflows, supporting text, voice, and visual inputs with real-time voice synthesis. The client supports both hosted Systemprompt MCP servers (configurable via a Systemprompt API key) and custom local MCP servers through a local configuration file. It is in early access and optimized for modern browsers (tested on Chrome) and developer workflows.