According to Hugging Face, the Reachy Mini conversation app can now utilize tools hosted in public Hugging Face Spaces via the Model Context Protocol (MCP). This allows users to give the robot new capabilities, such as web search or weather checking, by adding a Space from the Hub instead of editing the core application code. Because the tools continue running directly inside the Space, no code needs to be downloaded onto the local machine, and developers can publish their own tools for others to use. Previously, all tools were local, shipping inside the app and primarily handling the robot's physical body, such as head movements or playing emotions. Active tools are controlled by profiles containing an instructions prompt and a tools.txt file that acts as a gatekeeper. With the new implementation, remote MCP tools join built-in and local custom tools in a shared registry. The installation process validates the Hub Space, probes the MCP endpoint, discovers tools, and appends tool IDs to the active profile. Local aliases are generated to prevent naming collisions, and duplicate safety checks ensure unique tool names across the entire merged set. To test the feature, Hugging Face introduced canary tools and specialized profiles combining expressive physical actions with search and weather functions. The implementation highlights that Spaces must properly behave as MCP servers for tool discovery to succeed. Creators looking to share tools can publish them as public Gradio Spaces exposing standard MCP endpoints, keeping tools stateless and tagging them with reachy-mini-tool and mcp for discovery.
Adding MCP Tools to Reachy Mini
The Reachy Mini conversation app now supports tools hosted in public Hugging Face Spaces via the Model Context Protocol (MCP). Users can add capabilities like web search or weather checking without editing the app or downloading code locally. The system integrates these remote tools alongside built-in physical commands using profile configuration files.
Why it matters
This development decouples physical robot body controls from software capabilities like web search, enabling easier community contribution and modular updates for embodied AI systems without altering core application code.
Topic dossiers Hugging Face