Show HN: Clog – CLI and skill for debugging apps (github.com)

A new tool called Clog has been launched to enhance the debugging process for AI coding agents by facilitating runtime-driven debugging. Traditional debugging methods often falter when it comes to issues like race conditions or unexpected input shapes, as these cannot be fully understood by merely analyzing static code. Clog provides a local server that collects real-time logs from running applications, allowing AI agents to diagnose problems based on actual execution behavior rather than assumptions. The significance of Clog lies in its ability to streamline the debugging workflow. By running on a background server and collecting structured logs via simple HTTP POST requests, it enables the AI agent to track and reproduce bugs more effectively. Clog employs a structured five-phase process, where the agent interacts with the user to gather context, instrument the code with temporary logging calls, and then analyzes the timestamped logs to pinpoint discrepancies between expected and actual behavior. This approach not only improves debugging accuracy but also keeps the codebase clean by removing all instrumentation after the diagnosis. Clog is designed to be language-agnostic and integrates seamlessly with various codebases, making it a versatile tool for developers working with AI-generated code.