Cloud Calculator

This repository contains a cloud-deployable math tool server using the Model Context Protocol (MCP) via Server-Sent Events (SSE). This architecture allows students to connect their local local LLM agents (via VS Code Continue) to a centralized, instructor-managed tool.

Distributed Architecture

The following diagram shows how the system is distributed between the instructor's cloud infrastructure and the students' local machines.

graph TB
    subgraph "Cloud Infrastructure (Instructor)"
        VM[Cloud Virtual Machine]
        Docker[Docker Container]
        Uvicorn[Uvicorn / SSE Server]
        CalcTool[Calculator Logic]
        
        VM --> Docker
        Docker --> Uvicorn
        Uvicorn --> CalcTool
    end

    subgraph "Student Machine"
        VSCode[VS Code + Continue]
        LLM[Local LLM / Ollama]
        NB[01_math_agent.ipynb]
    end

    VSCode -- "HTTP-SSE / Port 8080" --> Uvicorn
    NB -- "Standard Logic" --> OpenAI[OpenAI / API]

Protocol Interaction Flow

When a student asks a math question, the following exchange occurs via the SSE protocol:

sequenceDiagram
    participant S as Student (VS Code)
    participant C as Cloud server (VM)
    participant T as Calculator Tool

    S->>C: GET /sse (Connect)
    C-->>S: SSE Stream Opened
    S->>C: POST /sse?sessionId=... (Initialize)
    C-->>S: JSON-RPC (Capabilites & ServerInfo)
    S->>C: POST /sse (tools/call: calculate)
    C->>T: eval(expression)
    T-->>C: Result (e.g., "42")
    C-->>S: JSON-RPC Result
    S->>S: Display "The answer is 42."

Student Setup (config.yaml)

To connect to the cloud tool, update your ~/.continue/config.yaml as follows:

mcpServers:
  - name: cloud-calc
    type: sse
    url: http://<YOUR-VM-PUBLIC-IP>:8080/sse

experimental:
  autoExecuteTools: true

Instructor Setup (Makefile)

Use the provided Makefile on your VM to manage the service easily:

Command	Action
make build	Builds the mcp-calculator Docker image.
make run	Starts the server in background (detached) on port 8080.
make logs	Follows the live server logs (useful for debugging student connections).
make stop	Gracefully stops and removes the running container.

VM Security Note:

Ensure your VM's Cloud Firewall (Security Group) allows Inbound TCP traffic on Port 8080.

Debug:

curl -v http://YOUR-VM-PUBLIC-IP>:8080/sse