Always-On Debugger

Automatically debug code from your terminal.

Overview

Always-On Debugger is a tool that enhances your terminal experience by automatically detecting errors and providing debugging assistance using AI. It acts as a wrapper around your existing terminal, intercepting commands and their outputs to offer real-time debugging support.

At present, we only support using Anthropic's Claude API. Let us know if you need OpenAI support.

https://github.com/user-attachments/assets/2a8a6d49-8b98-4319-a8de-16ef6ea23210

Loom Demo

Features

• Mimics the terminal for every command
• Automatically detects errors in command outputs
• Captures context and sends it to an AI language model for analysis
• Provides AI-generated debugging suggestions directly in the terminal

Installation

There are two ways to setup Always-On Debugger.

Option 1: Using npm

Step 1: Install the package

npm install -g aidebug

Step 2: Setup the API key for Anthropic

export ANTHROPIC_API_KEY=<PASTE_YOUR_OWN_API_KEY>

Step 3: Now you can use the debug command to debug your commands.

debug python average.py

Option 2: Manual Setup

Step 1: Clone the repo

cd ~
git clone [email protected]:samarthaggarwal/always-on-debugger.git

Step 2: Update ~/.bashrc Add the following to your ~/.bashrc or ~/.zshrc

alias debug="python ~/always-on-debugger/terminal.py"
export ANTHROPIC_API_KEY=<PASTE_YOUR_OWN_API_KEY>

Step 3: Source ~/.bashrc or ~/.zshrc . Alternatively, open a new terminal.

source ~/.bashrc

Usage

Just prefix any terminal command with debug. That's it, the debugger will automatically kick in when an error is detected and prints the error along with the suggested course of action. Here's an example:

Normally, the deverlop would only see the error.

[14:57:19] ➜  demo git:(main) ✗ python average.py
The average is: 3.0
Traceback (most recent call last):
  File "/Users/samarthaggarwal/personal/always-on-debugger/demo/average.py", line 15, in <module>
    average_of_empty = calculate_average(empty_list)
                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/samarthaggarwal/personal/always-on-debugger/demo/average.py", line 7, in calculate_average
    return total / count
           ~~~~~~^~~~~~~
ZeroDivisionError: division by zero

Prefixing the same command with debug prints the error along with diagnosis and recommendations.

[14:57:21] ➜  demo git:(main) ✗ debug python average.py
Traceback (most recent call last):
  File "/Users/samarthaggarwal/personal/always-on-debugger/demo/average.py", line 15, in <module>
    average_of_empty = calculate_average(empty_list)
                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/samarthaggarwal/personal/always-on-debugger/demo/average.py", line 7, in calculate_average
    return total / count
           ~~~~~~^~~~~~~
ZeroDivisionError: division by zero

---------------------------------------------------------------------
Debugging...

---------------------------------------------------------------------
To fix this error and improve the calculate_average function, follow these steps:

1. Modify the calculate_average function to handle empty lists:
   def calculate_average(numbers):
       if not numbers:  # Check if the list is empty
           return 0  # or you could return None, or raise a custom exception
       total = sum(numbers)
       count = len(numbers)
       return total / count

2. This modification checks if the input list is empty before performing any calculations. If it is empty, it returns 0 (or you could choose to return None or raise a custom exception, depending on how you want to handle this case).

3. Test the function with both non-empty and empty lists to ensure it works correctly in all cases.

4. If you want to keep the original loop structure, you can modify it like this:
   def calculate_average(numbers):
       total = 0
       count = 0
       for num in numbers:
           total += num
           count += 1
       if count == 0:
           return 0  # or None, or raise an exception
       return total / count

5. Choose the implementation that best fits your needs and coding style.


By implementing one of these solutions, you will prevent the ZeroDivisionError and handle empty lists gracefully.

How it works?

1. Prefix your terminal commands with debug.
2. If an error occurs, Always-On Debugger automatically captures the context.
3. The error context is sent to an AI language model for analysis.
4. Debugging suggestions are printed directly to your terminal.

Project Structure

• debugger.py: Core Python script that orchestrates the debugging flow
• llm.py: Handles LLM interaction (prompt generation and response parsing)
• (Additional files for terminal wrapping and packaging)

License

MIT License