Packrun is a package registry and comparison tool that helps developers and AI agents discover and evaluate npm packages across fifty-plus categories.
The tool addresses the problem of choosing between competing npm packages by automatically scoring them on a consistent formula that weighs downloads, bundle size, freshness, community engagement, and code quality. It surfaces real-time metrics including weekly download counts, gzip bundle sizes, maintenance activity, and security considerations. The scoring system assigns each package a 0-100 score within its category, making it possible to compare alternatives side by side and understand what distinguishes one option from another.
Developers should consider this tool when evaluating dependencies across common categories like HTTP clients, date libraries, validation, state management, ORMs, testing frameworks, and bundlers. It suits teams that want objective, formula-based comparisons rather than subjective recommendations, and it integrates directly into AI coding assistants like Cursor and Claude through an MCP server, allowing agents to search packages, check maintenance status, compare options, and validate compatibility before installation. The tool powers searches in under fifty milliseconds using Typesense Cloud.
The project is built as a monorepo using Turborepo, Next.js, Shadcn, and Tailwind. Development requires Bun, a Typesense Cloud account, and a Redis instance. The codebase is written in TypeScript and publicly available under the MIT license.
