Debba, a developer working on side projects, has released GapHunter โ€” a Claude Code skill that transforms negative competitor reviews into actionable product roadmap items. The tool automates competitive research by searching multiple platforms simultaneously and organizing the findings into prioritized implementation steps.

The Problem: Roadmapping Paralysis

According to debba's DEV.to post, deciding what to build next is the part of running a side project they struggle with most. "Open the issue tracker, stare at it, close the issue tracker," they wrote. "Repeat next weekend." Rather than continuing that cycle, debba decided to ask people who had already abandoned competitors โ€” specifically those who took the time to write detailed 1-star reviews explaining what they wanted and didn't get.

How GapHunter Works

Running the skill requires a simple command: /gaphunter DBeaver TablePlus (or any competing tools). The tool then searches G2, Capterra, TrustRadius, Reddit, GitHub Issues, and Hacker News in parallel. Once results are collected, it performs semantic clustering to group near-duplicate complaints โ€” so "no dark mode" and "lacks dark theme" collapse into a single finding rather than inflating the noise. The skill also reads your own repository structure: package.json, Cargo.toml, and the src/ tree. It cross-references competitor gaps against what you've already shipped, filtering out features you don't need to build because they're already implemented.

Output Format

GapHunter generates two files in a docs/ directory. The first is a JSON sidecar for programmatic access or tooling integration. The second is a self-contained HTML report that opens with a double-click โ€” no server required, no dependencies to install. The report features tabbed navigation covering Summary views, Quick Wins (high priority combined with small implementation effort), a Comparison matrix when analyzing multiple competitors, and a Plan tab listing specific implementation steps alongside the files in your repo that would need changes for each finding.

Current Limitations

Debba is upfront about what GapHunter can't do yet. External sources occasionally return 403 errors and become inaccessible. The semantic clustering still occasionally groups findings incorrectly. And perhaps most obviously, it cannot see private repositories belonging to competitors โ€” a fundamental constraint when analyzing companies that don't open-source their codebases.

Key Takeaways

  • Reviews are roadmap gold: Users who leave detailed negative reviews often articulate exactly what your product needs to win them over
  • Semantic clustering reduces noise: Grouping similar complaints prevents analysis paralysis and highlights real patterns
  • Codebase cross-referencing saves time: Knowing which gaps you've already filled prevents redundant work
  • MIT licensed for easy forking: The skill, HTML viewer template, and CSS are all available on GitHub under permissive licensing

The Bottom Line

GapHunter is a clever example of using AI to solve the eternal side-project problem โ€” not what to build, but how to prioritize when everything feels urgent. It's not a magic bullet (those 403 errors and imperfect clustering will frustrate power users), but for solo devs or small teams who want data-driven roadmap decisions without spending weeks on manual research, it's worth installing. The entire project was built inside Claude Code itself, which is either a flex or an admission that even developers enjoy automating their own workflows.