KitGrade: starter-kit scores from public evidence

kitgrade.thecompound.tech

KitGrade scores SaaS starter kits from public evidence, using a Next.js application and a captured read API. Developers use it to compare maintenance, completeness, transparency, documentation, and support before choosing a starting point.

The problem

Starter-kit directories usually put names, stacks, prices, and feature badges beside each other. That helps with discovery, but it does not show whether a repository builds cleanly or whether a claim came from a file anyone can inspect.

KitGrade keeps those facts separate. A score can cite a commit history, dependency manifest, pricing page, documentation index, or repository path, and the product labels evidence it could not measure instead of treating it as zero.

How KitGrade works

KitGrade keeps a scorecard with five components worth a full hundred points. The data layer reads four captured JSON files, and the capture refuses to write a set whose component figures do not add back to the stored totals.

1. The capture script reads the public API for statistics, method rules, the ranked kit list, and each kit detail record.
2. The score resolves maintenance, completeness, transparency, documentation, and support from the stored evidence fields.
3. The product records whether evidence came from a clean checkout or from reading the repository, manifest, and documentation.
4. The Next.js routes expose the same data through /api/stats, /api/method, /api/kits, /api/kits/<slug>, and /api/compare.
5. The console draws measured components as filled spans and unmeasurable components as dotted caution tracks.

The distinction between "We ran it" and "Read, not run" is part of every score. A clean-checkout run keeps every command, exit status, and duration with the kit.

A worked example

A kit detail record can include its homepage, repository, stack, licence, price evidence, score breakdown, fourteen capability signals, and an install transcript. The transcript records commands such as cloning a repository and running pnpm install --frozen-lockfile, with the result and duration attached.

The product also keeps an explicit boundary. If a closed kit has no public repository, KitGrade marks maintenance as unmeasurable and keeps those points in the denominator.

How it compares

Boilerplates Directory compares SaaS starter kits by stack and uses open-source and price badges on listings. It is a better first stop when the job is broad discovery across stacks.

BoilerplateList lists boilerplates and starter kits with stack, pricing, and licence information on detail pages. It fits a reader who wants catalogue metadata before opening a repository.

BoilerplateHub compares boilerplates through feature filters such as payments, authentication, databases, documentation, and support. Its deals page also lists pricing from free options to premium kits.

KitGrade differs by attaching a published rule and evidence source to each scored component, then separating clean-checkout runs from documented-only records. The result is a scorecard about what the method could verify, not only what a listing claims.

Limits

KitGrade does not infer evidence that a public surface does not expose. It also does not turn a failed build into a lower score, because the method records the run result separately from the score.

A reader who needs a feature catalogue should use BoilerplateHub or Boilerplates Directory. A reader who needs repository evidence and a reproducible scoring method should use KitGrade.

Try it

The first step is to put a kit on the bench. KitGrade then shows the score, the evidence behind each component, and the parts the method could not measure.

kitgrade.thecompound.tech