IntelliFSharp

🚀 IntelliFSharp - F# Code Runner with LLM Feedback Loop

An AI-powered playground where your F# code gets automatically tested and fixed! This project uses Large Language Models (Ollama) to generate F# code, executes it in real-time, and sends errors back to the AI for auto-correction 🔄

"It's like having a pair programmer who never sleeps! 😴➡️🤖"

🌟 Key Features

• AI-Powered Code Generation - Uses Ollama LLMs to write F# code
• Real-Time Execution - Runs code in F# Interactive (FSI) sandbox
• Auto-Debugger - Sends errors back to AI for automatic fixes
• Smart Feedback Loop - Up to 3 correction attempts
• Emoji-Powered Logging - Because debugging should be fun 🐛➡️✅

⚙️ How It Works

sequenceDiagram
    participant User
    participant Program
    participant Ollama
    participant FSI
    
    User->>Program: Start program
    Program->>Ollama: Send system message
    Ollama-->>Program: Acknowledge
    Program->>Ollama: Send coding task
    Ollama-->>Program: Generated F# code
    Program->>FSI: Execute code
    alt Success
        FSI-->>Program: Results
        Program-->>User: 🎉 Success + Output
    else Error
        FSI-->>Program: Error details
        Program->>Ollama: Send errors
        Ollama-->>Program: Fixed code
        loop Max 3 attempts
            Program->>FSI: Retry fixed code
        end
    end

Loading

🚀 Getting Started

Prerequisites

• .NET 6+ SDK
• Ollama (with at least one LLM installed)
• F# Interactive (dotnet tool install -g fsi)

Installation

1. Clone the repository:

git clone https://github.com/pwrmind/IntelliFSharp.git
cd IntelliFSharp

2. Run the project:

dotnet run

Configuration

Modify these constants in Program.cs:

const string modelName = "deepseek-coder-v2:latest"; // Change to your preferred model
const int maxAttempts = 3; // Max auto-fix attempts

🧪 Example Run

📩 Sending system message...
✅ System message sent

📤 Sending task:
📝 Task: Write F# function calculating sum of squares from 1 to n

📥 LLM response:
Here's the F# code:
```fsharp
let sumOfSquares n = 
    [1..n] |> List.map (fun x -> x * x) |> List.sum

sumOfSquares 5;;

⚡ Executing block #1... ✅ stdout: 55 ❗ stderr:

🎉 Code executed successfully! 📋 Results: 📦 Block #1 result: 55 ══════════════════════

## 🛠️ Customization Tips
1. **Change Tasks** - Modify the `taskMsg` variable for different challenges:
```csharp
var taskMsg = "📝 Task: Write F# function to calculate Fibonacci sequence";

2. Add More Models - Try different Ollama models:

const string modelName = "codellama:7b"; // Smaller/faster model

3. Extend Timeouts - For complex tasks:

bool completed = outputCompleted.WaitOne(30000); // 30s timeout

🤔 Why This Is Cool

• Perfect for Learning F# - See how AI solves programming challenges
• Debugging Assistant - Watch how LLMs fix their own mistakes
• Prototyping Tool - Quickly test algorithmic ideas
• LLM Testing Platform - Evaluate code generation quality

🚧 Known Limitations

• Requires running Ollama instance
• Simple error handling (WIP)
• FSI state persists between runs
• Currently supports single-file programs

🌈 Future Improvements

• Add support for multiple files
• Implement FSI session reset
• Add GUI interface
• Support other languages (Python, JavaScript)
• Add benchmark mode

---

Made with ❤️ and too much coffee ☕ by [Your Name Here]. Contribute and make it better!