Seance Stock

You know what's tragic? Millions of abandoned projects on GitHub. Brilliant ideas, left to die.Seance Stock finds them, and tells you how to bring them back to life.Let me show you 👻

Comment

Inspiration

Every developer has been there: you're searching for a solution, you find the perfect library on GitHub, and then you see it—last commit: 4 years ago. Dependencies outdated. Issues piling up. A digital graveyard.

The tragedy isn't that projects get abandoned. Maintainers move on, priorities shift, life happens. The tragedy is that we lose incredible work simply because no one knows where to start bringing it back.

I built I built Seance Stock for the Kiroween Resurrection Track because the theme hit home: great code deserves a second life. I wanted to make resurrection effortless 😈

What it does

Three core functionalities:

  1. Smart Discovery Search GitHub for projects that have been abandoned for 2+ years but still have community interest (10+ stars). Filter by programming language to find projects in your expertise area.

  2. Dual Scoring System Every repository gets two scores:

Abandonment Score (0-100): How "dead" is this project? Measures years of inactivity Counts unresolved issues Evaluates community engagement Revival Potential (0-100): Is it worth saving? Weighs star count and community interest Analyzes fork activity Considers open issues as demand signals Identifies tech stack viability

  1. Resurrection Plans Get a customized, language-specific modernization roadmap for each project: Dependency updates tailored to how long it's been abandoned Modern tooling migrations (e.g., Vite for JavaScript, pyproject.toml for Python) Testing framework recommendations Issue triage strategies CI/CD setup guidance Community outreach steps

How we built it

Built entirely with Kiro AI and testing on Vercel. I used Kiro super skills in the following: This project showcases Kiro's full capabilities: Agent Hooks: Auto-lint on save, API change notifications Steering Docs: Consistent vintage theme and code style Spec-Driven: Structured requirements for scoring algorithms MCP: GitHub API integration for real-time data

Challenges we ran into

Github API rate limi and Balancing Scoring Algorithms. Balancing the score was a little hard as early versions of my abandonment and revival scores were unbalanced. Popular projects with thousands of stars would score high even if they were truly dead, while niche but actively-needed projects scored low.

Why It Happened: I initially over-weighted star count without considering the relationship between popularity and actual revival viability.

The Solution: Iteratively tuned weights through testing with real repositories: Capped star contribution at 40 points (not 100) Added fork activity as a separate signal Weighted open issues as "demand" rather than just "problems" Found the sweet spot: 30% inactivity, 30% issues, 40% community engagement

Accomplishments that we're proud of

Getting this up and running as a side project with Kiro was a blast ⚡️✨

What we learned

  1. Real GitHub Integration This isn't a mockup or demo with fake data. Seance Stock queries live GitHub repositories, analyzes real metrics, and surfaces actual abandoned projects that developers can fork and revive today.

  2. By defining our design system upfront in .kiro/steering/, every component Kiro generated automatically matched our aesthetic. No manual CSS tweaking, no style inconsistencies—just coherent, beautiful output from the first generation.

  3. Agent Hooks Eliminate Context Switching: Having ESLint and TypeScript checks run automatically on save meant we caught errors instantly, without breaking flow to run terminal commands. This small automation had a massive impact on development speed.

What's next for Seance Stock

Now I am excited to turn this into a full app. Will work more on it and hope it is used by the Open Source community.

Built With

Share this project:

Updates