The goal is to solve a problem I kept running into: there are thousands of MCP (Model Context Protocol) servers out there, but no easy way to know which ones are actually good, reliable, or worth your time. Many MCP repositories exist but the evaluation of the servers - when present - is managed by the administrator of the repo; reviews and tales of experience by real users are then scattered on Reddit or Discord, making the experience of building AI workflows long and complicated.
In MCP Review, anyone can review and rate servers, helping others in the choice of the best tool for their needs. Reviews and rates are tied to a GitHub account, so we prevent abuse and visitors can verify the competence of the reviewer. At the moment, I built the infrastructure's main feature and I am encouraging developers to upload their work and join the community. It takes less than 2 mins if a public Github repo is available and the Server card is highly customisable, from description to tool list to custom icon.
The app is built with - Next.js 16 (App Router) + TypeScript - PostgreSQL + Prisma - NextAuth.js (GitHub) - Tailwind CSS + Radix UI And it is fully open-source, any contribution is welcome (https://github.com/ggange/mcp-review). You can browse servers without signing in. Authentication (GitHub) is only needed to rate or upload servers.
I am continuously working and I'd love feedback on the rating/review system, search and filtering UX, what features would make this more useful and any bugs or issues you encounter.
Thanks for checking it out!