ChatDBG emerges as an AI-powered debugging assistant, enhancing the programmer’s toolkit by facilitating a dialogue-based interaction with the debugger ChatDBG Assistant. It autonomously navigates stacks and inspects program state to troubleshoot issues like crashes or assertion failures. The combination of LLMs with debuggers allows more natural inquiries such as ‘why is x null?’, enhancing the debugging process. The ChatDBG tool integrates with LLDB, GDB, WinDBG for native code, and Pdb for Python, demonstrating significant improvements in automated program repair with an 85% success rate, following up on queries to fix bugs.

• ChatDBG marries LLMs with traditional debuggers for enhanced debugging.
• It allows programmers to ask complex questions about a program’s state.
• ChatDBG can autonomously inspect and interact with program state.
• Achieves high success rates in generating fixes for a range of errors.
• Compatible with major debuggers and has been downloaded nearly 30,000 times.

The introduction of ChatDBG marks a significant advance in debugging tools, reflecting the potential of AI to revolutionize traditional software development methodologies. By augmenting human intelligence with machine precision, ChatDBG offers faster debugging and a more intuitive user experience.