PraxisAI: Practice AI coding interviews on real GitHub issues

Hey Product Hunt! More and more coding interviews now let you use an AI agent, and what gets judged is how well you work with it on a real codebase. But there was nowhere to actually practice that format. LeetCode puzzles don't test it. So we built PraxisAI: - Pick a real GitHub issue from SWE-bench (Django, Flask, SymPy, matplotlib...), checked out at the commit where it was still broken - Fix it in browser VS Code with an AI agent, a timer and a token budget - Hidden tests grade your fix. You get a score, verdict and test outcome for free Paid plans add a detailed report across 6 scored dimensions, a prompt coach that flags habits like raw-error pastes and vague retries, and a replay of every prompt and agent action. There are also 24-hour take-home projects and a leaderboard. Free tier: 2 full-model rounds a week, no card needed. We'd love your feedback, especially on the report: what would you want to see after a round?