sakethkanchi/resume-radar

Many companies now use AI-powered tools to screen resumes before a human ever reads them. These tools do not just scan for keywords. They fetch the candidate's live GitHub profile, categorize each repository, and apply a scoring rubric across several categories. Resume Radar is a Claude skill that teaches the AI assistant to apply the same scoring logic, based on a rubric reverse-engineered from HackerRank's open-source hiring agent. When you trigger the skill, Claude runs a Python script that calls the GitHub API and classifies your public repositories the same way the screener does, separating open-source contributions from self-built personal projects. That classification matters because some scoring categories have caps: if all your repos are personal projects with no merged contributions to external codebases, the open-source score is capped at a much lower ceiling. After enriching the GitHub data, Claude scores your resume across four weighted categories plus bonuses and deductions, then returns a table sorted by how many points each gap is costing you, alongside specific fixes. For example it might note that adding a live demo link to each project would recover several points, or that contributing a merged pull request to a well-starred open-source repo would unlock the largest gain. The skill is explicit that it will only suggest things you genuinely have or can do. It does not suggest fabricating experience, inventing metrics, or claiming contributions that did not happen. Its position is that the screener reads the live repo and a human reads the PDF, so dishonest signals get caught anyway. To use it, you drop the skill folder into Claude's skills directory and ask Claude to audit, simulate, or optimize a resume. Setting a GitHub API token raises the rate limit from 60 requests per hour to 5000. The project is licensed under MIT.