The standard

How it is checked.

Every build is checked before it is listed, and every listing says how.

What we check before a build is listed

Before a build is listed we confirm that the file opens cleanly, that it matches what its listing says, and that its guide can be followed.

We look for keys left in and for data sent somewhere the listing does not mention.

The two levels

The listing shows the level the check reached.

  1. File and guide read

    Before a builder's build is listed we open its file and read it and its guide against every claim in the listing.

  2. Run on sample data: FHQ Tested

    We run a build only when its builder asks for a run, and only for a tool we can run, which is n8n today. When we have run one on our own test accounts with sample data, its listing carries the mark "FHQ Tested".

A build with the mark was also run on sample data.

Who and what does the checking

Automatic tests run on the file first, then an AI assistant reads the file and its guide. We look over a sample of the results ourselves, and every refusal a builder disputes.

The check is proportionate and stops short of a full audit, and each listing says what was done.

What the check does not cover

Our check leaves four things out:

  • a test with your own data, accounts or settings;
  • changes made after the check to the tools the build runs on;
  • a full security audit;
  • any promise that the build suits your business or will keep working.

Read the guide and the listing's limits before you use a build, and test it on sample data first.

A check gives a sensible estimate of quality and stops short of a full audit.

The check is our opinion on the day. It does not make us responsible for the build.

When a build fails

A build that fails is not listed.

We can refuse a build, and we give you the reasons. If we ask for changes, you can send the build again once at no cost.

How long it takes

Every build stays in review for at least 24 hours, and nothing is approved faster.

Join the list