Inspiration
Modern software incidents generate overwhelming amounts of logs, stack traces, and fragmented context. Engineers often spend more time gathering evidence than solving the actual problem. We built Traceback Agent to streamline this workflow into a guided AI-assisted investigation that reduces time from detection to resolution.
What it does
Traceback Agent provides an end-to-end workflow for production incident response. Users can create incidents, analyze evidence, organize hypotheses, generate AI-assisted code patches, validate proposed fixes, and export detailed investigation reports. Instead of jumping between multiple tools, engineers work through a single structured investigation pipeline.
How we built it
We built the application using Next.js and TypeScript with a modern React interface. OpenAI Codex accelerated implementation, refactoring, testing, and UI improvements throughout development. GPT-5.6 was used to help design the investigation workflow, review architecture, improve user experience, and refine AI-assisted reasoning throughout the project.
Challenges we ran into
Designing a workflow that felt useful rather than overwhelming was one of the biggest challenges. We focused on making each investigation stage clear while keeping the interface responsive and intuitive. Another challenge was integrating AI assistance without interrupting the investigator's decision-making process.
Accomplishments that we're proud of
Built a complete end-to-end incident investigation workflow. Created an intuitive interface for structured debugging. Integrated AI throughout multiple investigation stages. Produced automated investigation reports. Delivered a polished production-style application during Build week
What we learned
I have learnt codex ,codex skills its functions and different ways to build an end to end product with codes
What's next for Traceback
real patch generation, automated CI/CD validation, support for multiple LLM providers, collaborative investigations, enterprise authentication, and integrations with platforms such as GitHub, Jira, Slack, and observability tools.
Log in or sign up for Devpost to join the conversation.