How we score tools
Every tool page shows a number out of 100. This page explains exactly where it comes from: every input, every weight. No black box, no pay-to-play. Most inputs are public GitHub data you can verify yourself. A few come from our own tracking, and we say so below: Reddit chatter, whether we have featured a tool, and whether we have written its Lens yet.
One thing before the details: a low score is not a verdict on quality. It usually means a tool is young, niche, or already so established that our momentum math deliberately ignores it. The score gets a tool on your radar. The Lens tells you whether to commit.
There are two scores
The Score on each tool page is a report card: how adopted, maintained, and trusted a project is today. The Discovery Score powers our momentum lists and answers a different question: what is gaining traction right now? A tool can ace one and flunk the other. That is by design.
The Score: a report card out of 100
Five dimensions, weighted by how much they should matter when you are deciding whether to depend on something.
GitHub stars and forks, on a curve. Maxing it out takes six-figure stars, or 50k with a serious fork base. The curve is steep at the bottom: a young tool with 400 stars starts low here no matter how good it is.
How recently code actually moved. We measure the last push to the repo: active this week scores the full 25, quiet for six months sits in the middle, untouched for over a year scores zero. A repo we have not measured yet sits at a neutral 10, not zero.
Trust signals (organization backing, a notable author, an active fork base, GitHub Sponsors) plus whether developers are actually discussing it on Reddit.
Permissive licenses (MIT, Apache) score highest because they ask the least of you before you ship. Copyleft scores lower, source-available lowest. This is about friction, not ideology.
Whether we have written a Lens for the tool yet. If these points are missing, that is on us, not the tool.
Grades follow the number: 90+ is an A+, 80 an A, 70 a B+, 60 a B, and so on down. A "C" tool is not a bad tool. It is usually a young one that has not accumulated stars and Reddit chatter yet.
The Discovery Score: what is rising
Recomputed daily. You will not see this number printed anywhere: it works behind the scenes, ranking the catalog to decide what surfaces in our momentum lists. Built to catch tools gaining traction before everyone knows about them. Seven components:
Percentage growth, not raw stars. A repo going from 200 to 2,000 stars beats one sitting at 50k. This is the core of the whole system. Tools that share a monorepo's stars get this halved, so one famous repo does not inflate dozens of listings.
Newer repos score higher. This decays as the project ages, on purpose.
Is it growing faster right now than its own recent average? Rewards tools that are picking up speed, not just moving.
Releases shipped in the last 30 days. Shipping counts. Honest caveat: our release tracking is behind right now, so this component under-counts until we fix the plumbing.
Tools we have never featured in the newsletter get a bonus. We are trying to show you things you have not seen.
Fork-to-star ratio. People forking a repo are using it, not just bookmarking it.
Tools above 50k stars lose points, scaling with size. You already know about Kubernetes. That is not what this score is for.
New tools need about a week of velocity history before this number settles. If a tool just landed in the directory, its Discovery Score will run low until we have real data. That is measurement lag, not a judgment.
Why a low score is not a bad tool
The Discovery Score punishes being big and established. Kubernetes will never top our momentum lists, and it should not. The report-card Score leans on public signals: stars, releases, license, community chatter. A sharp niche tool with 400 stars and no Reddit thread can sit in the C range, and it might still be exactly what your project needs.
No score can see code quality, documentation, or whether a tool fits your stack. That is what the Lens is for: a human analysis of what the tool does well, what it costs you, and what the catch is.
What we will not do
Nobody can buy a point. No tool pays to be listed, featured, or scored. If you maintain a tool and think we got something wrong, email us and we will look at the data together.