How we test

The rules below are the product. Everything else on this site is their output.

1. We actually run the thing

No page ships until the skill has been installed in a clean environment and exercised against a minimal real task — the scanner scanned something, the renderer rendered something, the CLI returned real output. Reading a README is research; it is not a test.

2. The environment is disclosed, always

Every test record states the OS, Node, npm and Python versions it ran on, and the exact upstream commit tested. Our default box is a headless Linux sandbox with two cores — closer to CI and to cheap servers than to a maxed-out MacBook, which is why we catch failures that Mac-based reviews never see. When a skill needs a desktop environment we say precisely where our verification stopped, rather than pretending it went further.

3. Timings are measured, not estimated

Install times come from wall-clock measurement around the actual command. One-off costs (CLI bootstrap, dependency install) are reported separately from the skill's own install so you can predict your second install, not just your first.

4. Failures are content

Every failure is recorded verbatim — the step, the error, and the fix if we found one, as a three-part entry in the test log. A skill with zero failures gets that stated explicitly, because "nothing broke" is information too. We do not penalize a skill for our environment's constraints; we document the boundary honestly and move on.

5. Numbers are computed, never typed

Test records live as raw JSON logs. Every statistic on this site — install times, failure counts, star counts with their as-of dates, the summary table on the front page — is generated from those logs by the site build script. If a number cannot be traced to a log entry, it does not appear. Star counts drift daily and are therefore always stamped with the date we read them.

6. Retests are scheduled, not aspirational

Skills ecosystems move weekly. Each tested skill goes on a 30-day retest calendar; every retest appends to the changelog with anything that changed — including upstream repos that moved or died. The "tested on" date on every page tells you exactly how fresh the record is.

What we are not

Not a directory (we link several good ones from individual pages when relevant), not affiliated with Anthropic or with any skill author, and not paid for placement. If a skill author disputes a finding, we retest and publish the correction in the changelog — corrections are content here, not embarrassments.