Hare has no package manager by design, so third-party modules are managed either via system-wide installation using `make install` or by vendoring directly into a project with `git subtree`. The post explains how Hare's module resolution works via HAREPATH, the standard repo layout convention (hare-MODULE with module code in a top-level directory), and walks through both approaches with concrete Makefile examples. A key gotcha is that setting HAREPATH as an environment variable completely replaces the compiled-in default, requiring explicit re-inclusion of stdlib paths. The `vendor/*/` glob convention mentioned in Hare docs does not actually work, so each vendored path must be listed explicitly. The recommendation is to use system-wide install for personal/small projects and git subtree vendoring for collaborative or reproducible builds.
