Stars can be a useful discovery signal. They are also easy to overread.
A star does not tell you whether a project still ships releases, whether one exhausted maintainer owns every critical path, whether your migration can be reversed, or whether the documentation covers the version you will deploy.
When I evaluate a project, I would rather see a small set of understandable evidence:
- Recent releases tied to meaningful changes
- Maintainers reviewing outside contributions
- Issues that move through a visible lifecycle
- Clear support and security policies
- Reproducible installation and upgrade instructions
- A license compatible with the intended use
- Evidence from a workflow similar to mine
Even these signals can conflict. A mature database may have fewer commits because it is stable. A fast-moving AI tool may look active while breaking compatibility every week. Context matters.
AtlasRepo is exploring how to make those trade-offs visible without pretending one score can answer every adoption question. Browse the catalog and pick a repository you know well: what evidence would you show a skeptical teammate?
Which signal do you trust most — and which popular metric do you think is routinely misused?