Hey Product Hunt 👋 Replokit started from a problem every growing team hits: nobody can answer "what are we actually running?" Our services were scattered across multiple clouds, a handful of databases, and a pile of AI providers — spread over dozens of repos and email accounts. Worse, old projects we'd "shut down" were still wired to live infrastructure and quietly billing us every month. Existing tools wanted deep cloud access, agents, or a heavy catalog you have to maintain by hand. We wanted the opposite: point it at GitHub, get the whole picture in minutes, and trust it with the least possible access. So Replokit is read-only and GitHub-only — it reads your dependency, env, and config files (never your source, never secret values) to deterministically map every cloud, database, and AI model you depend on. Then it flags the things that hurt: forgotten infra still billing you, exposed secrets, vulnerable dependencies, and an estimated monthly spend. A Claude-powered summary turns it into a CEO/CTO-ready read, and it keeps watching — alerting you when something changes. The hardest part was earning trust: keeping it read-only, encrypting everything, and making sure your code is never stored. Would love your feedback — especially on what you'd want it to catch next. 🙏

Replokit

Map your stack from GitHub. Spot forgotten infra & cost

Map your stack from GitHub. Spot forgotten infra & cost