How we test templates
Every template is checked before it goes up, and shows its version and the date of its last check.
What we check
The file holds together
Each template is one source file: its inputs, its steps, its built-in checks and its outputs. A script reads the file and refuses it when a piece is missing or when something points at a part that is not there.
All four formats build
From that one file we build the prompt chain, the checklist, the Claude skill and the n8n file, and check that each is complete and says the same thing as the source. The n8n file is checked against the n8n import format.
The wording holds up
We read each template through for plain words, no invented facts and no figures presented as real. Every example in a template is made up for illustration and is labeled "Sample".
It is tried in a tool
For each AI tool we list, we run the template’s steps in it with the sample input and look at what comes back: does each step do its one job, and does each check catch what it is meant to catch.
What "works with" means
A tool is listed on a template only after we have run that template in it, or, for n8n, checked the file against the import format. The page shows the tool, which version of it we used and the date. A template with no tool listed has not been tried in one yet.
Versions and dates
Every template page shows its version and when it was last checked. The version goes up when the steps, checks or inputs change, not for a typo. The changes page lists every version of every template and every correction after publishing.
What we do not test
An AI tool can give a different answer each time you run the same steps, and its tools change without notice. We cannot promise the same result twice. That is why the checks sit after the steps, and why you should read what comes out before you use it.