MCP Server New!
The Diffusion MCP server lets you harness the strengths of modern generative AI in order to rapidly accelerate design and development of your real-time applications using Diffusion.
By integrating with your preferred LLMs (such as Claude, GPT and Gemini) and IDEs (including VS Code and JetBrains IDEs), the Diffusion MCP server provides all the necessary tools and context for your generative AI assistant to work with data in Diffusion.
Capabilities
Tasks which the tool can help with include:
- Analysing the topic structure of your Diffusion server
- Creating data transformations (topic views) using natural language
- Populating your Diffusion topic tree with dummy data for testing purposes
- Developing and testing Diffusion integrations with your application
See below for detailed example prompts.
WARNING
While generative AI tools are improving rapidly, the potential for LLMs to hallucinate or provide incorrect answers remains.
For best results, we recommend using the latest iterations of Anthropic or OpenAI models.
Video demo
Example prompts
- Connect to the Diffusion server at localhost:8080 with the credentials "admin" and "password".
- Please describe the configured topic views
- Please create a topic view called "source-to-json" which converts topics under the "source/" path to JSON
- Please create a metric alert that alerts if the diffusion_topics_current metric is above 5 please
- Please remove the above metric alert
- Create a topic of each supported type
If you are experiencing issues with these prompts, please make sure you are using an up to date model with good reasoning and tool-calling abilities.
Downloads
diffusion-mcp-server on GitHub 
Download latest releases of the Diffusion MCP server from GitHub Video tutorials
These in-depth video tutorials guide you through the use of Diffusion with the MCP server and Claude Desktop.