CrashCart: An AI Debugger That Has to Show Its Receipts (Built on Gemma 4)
This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle What I Built You run docker compose up . The API container dies. You get forty lines of noise, paste them into a chatbot, and get back a confident paragraph — sometimes quoting an error line that was never in your log. CrashCart is a debugging assistant that has to show its receipts. You paste a terminal log,…
CrashCart is an AI-powered debugging tool that can show its work, built on Gemma 4, an open-weight model. Users paste terminal logs, stack traces, or build failures into the tool, which then redacts any sensitive information like API keys and passwords. It uses Gemma 4 to generate a structured diagnosis, cross-referencing each line the model quotes with the original log to ensure accuracy.
If the model's claims don't match the log, CrashCart lowers its confidence rating and highlights the discrepancy. The diagnosis includes the smallest safe fix, step-by-step verification, and a ready-to-paste GitHub issue. CrashCart is designed to be a debugging assistant for freelance web developers who frequently debug client projects and need to explain failures to clients or colleagues.
The tool is free to use for a limited number of analyses per day or costs around $6/month for higher limits and client-ready reports. It is open-source and MIT-licensed, allowing anyone to use, modify, and contribute to the project. The code is available on GitHub, and the project was created as an entry for the DEV Hacktoberfest Open-Source AI Challenge.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
