VibeTracker

Inspiration

Vibe coding makes software development faster and more accessible, but it can also make projects difficult to understand and manage.

When developers use AI coding tools, a large amount of code may be generated in a short time. However, it is often unclear which features have been completed, which requirements are still missing, whether the frontend and backend are aligned, or whether the project is actually ready to ship.

We built VibeTracker to create a clearer connection between the original product idea, the implementation plan, and the current state of the codebase.

What it does

VibeTracker is a project tracking and monitoring tool designed specifically for vibe coding workflows.

It analyzes a software project and automatically generates an HTML-based visual dashboard that shows the current development status.

The dashboard can display:

Project goals and requirements Features and implementation tasks Completed, in-progress, blocked, and pending work Frontend pages and UI components Backend APIs and services Database models and integrations Authentication and permission systems Testing and deployment readiness Potential gaps between the original requirements and the actual implementation

Instead of reading multiple planning documents, code files, and AI chat histories, users can open one visual dashboard to understand what has been built, what is missing, and what should happen next.

How we built it

We designed VibeTracker as a lightweight project intelligence layer that works alongside AI coding tools.

The system reads project requirements, task definitions, source-code structure, and implementation signals from the repository. It then converts this information into structured project data.

That data is used to automatically generate a standalone HTML dashboard. The dashboard presents the project as visual modules, progress indicators, task lists, implementation maps, and status summaries.

We also designed the project around a workflow in which the AI first creates a structured implementation plan, then updates the tracking data as development continues.

The core workflow is:

Understand the product requirements. Break the project into features, modules, and technical tasks. Generate a structured project plan. Observe changes in the codebase. Compare implementation progress with the original plan. Generate or update the HTML dashboard. Highlight missing features, blockers, and inconsistencies.

The project can be integrated into a VS Code-based workflow and used together with tools such as Codex, Claude Code, and other AI coding agents.

Challenges we ran into

One of the biggest challenges was determining the real status of a feature.

A file may exist without the feature being functional. A page may be implemented without being connected to an API. An API may exist without database integration, authentication, error handling, or tests.

This meant that VibeTracker could not simply count files or completed tasks. It needed to evaluate multiple implementation signals and represent uncertainty clearly.

Another challenge was keeping the dashboard synchronized with a rapidly changing codebase. Vibe coding projects can change significantly after a single AI prompt, so the tracking system must update without requiring developers to manually maintain another project-management tool.

We also had to find a balance between structure and flexibility. Different projects use different frameworks, naming conventions, architectures, and development workflows. VibeTracker therefore needs to understand projects without forcing every developer to follow the same rigid template.

Accomplishments that we're proud of

We are proud that VibeTracker turns an abstract and often chaotic AI-assisted development process into something visible and understandable.

Instead of presenting progress as a simple percentage, it connects product requirements with actual implementation areas such as pages, APIs, databases, permissions, tests, and deployment.

We also created a dashboard that can be generated as a portable HTML file. It does not require users to configure a complex project-management platform, and it can be opened locally, shared with collaborators, or stored directly inside the repository.

Most importantly, VibeTracker is not intended to replace AI coding tools. It acts as a control and alignment layer around them, helping users understand whether the AI is building the right product—not just generating more code.

What we learned

We learned that the main challenge in vibe coding is not always code generation. It is maintaining alignment between user intent, project requirements, implementation decisions, and the actual state of the codebase.

AI coding agents are very good at completing local tasks, but they may lose awareness of the overall product. As a project grows, developers need an external system that preserves structure and makes progress observable.

We also learned that project status is multidimensional. A feature should not be considered complete simply because its UI exists. Completion may also depend on business logic, backend connectivity, data persistence, permissions, error handling, testing, and documentation.

This led us to think of VibeTracker not only as a progress dashboard, but as a project observability system for AI-assisted software development.

What's next for VibeTracker

The next step is to make VibeTracker more automatic and more deeply integrated into the development workflow.

We plan to add:

Real-time project change monitoring Automatic dashboard updates after code changes Deeper Git and commit-history analysis Requirement-to-code traceability AI-generated risk and blocker detection Frontend, backend, database, and testing health checks Support for more frameworks and project structures Integration with GitHub issues and pull requests Team collaboration and shared project views Historical progress tracking Automatic recommendations for the next development task

Our long-term goal is to build a control layer for vibe coding—one that helps developers, founders, and non-technical builders maintain visibility, alignment, and confidence while working with AI coding agents.

Built With

• api
• fast
• react