Feature Request: Add Model Context Protocol (MCP) support for AI-driven infrastructure orchestration

[klaveriuzent]

Problem / Context

Torollo is an amazing visual playground for backend engineering. It makes system design highly interactive and tangible by translating a visual canvas into real local Docker containers.

With the rapid growth of AI assistants (like Claude, ChatGPT, etc.) using the Model Context Protocol (MCP), engineers are increasingly using AI to help design, debug, and understand architectures. Currently, there is no native way for an AI agent to interact with or understand the Torollo workspace.

Proposed Feature

I would like to suggest adding native MCP Server capabilities to Torollo. By exposing Torollo's backend functions as MCP Tools, users can pair Torollo with an AI assistant to create a hybrid "AI + Visual" learning sandbox.

Here are a few potential use cases for Torollo MCP tools:

1. get_current_topology: Allows the AI to read the current canvas JSON structure so it can analyze the architecture and give feedback or recommendations.
2. add_node_to_canvas: Allows the AI to programmatically inject nodes (e.g., adding an Nginx Load Balancer or a PostgreSQL replica) based on user prompts.
3. simulate_traffic: Allows the AI to trigger pings or mock traffic between components to test security group/iptables rules.

Why this is highly valuable for Torollo:

• Next-Gen Learning Experience: Students can ask an AI agent, "Hey, my frontend container can't talk to the database, can you fix it?" The AI could inspect the visual canvas via MCP, modify the Security Groups, and the user would see the connection wire light up in real-time.
• AI-Driven Infrastructure Prototyping: It aligns perfectly with your upcoming "Terraform Generation" roadmap, allowing AI to help build the visual architecture before exporting it to IaC.

Thank you for creating this incredible tool! I would love to hear your thoughts on this.

[Derssa]

Hi! Thanks so much for this amazing suggestion.
We actually already have a backlog item/story for adding native MCP server support to Torollo's roadmap, and we plan to implement it in an upcoming version!
Your suggestions for tools like get_current_topology and add_node_to_canvas align perfectly with our vision. Adding MCP capabilities will be highly valuable not only for students, but also for senior/experienced engineers who want to use AI agents to prototype, visualize, and test system designs much faster.
We'll update this issue as we begin development. Thanks again for your support and feedback!