On that same page is Gherkin which is what both examples are. (I may have gotten the syntax slightly wrong.) Cucumber uses Gherkin. I forget which is which exactly. Maybe Cucumber is the code that reads Gherkin and executes step definitions.

For whatever reason, people try to make a small number of extremely flexible step definitions which leads to scenarios that don’t actually read as business requirements. (Which is the entire point in the first place.)

Given a user has a bank accout
And the account has a balance of $10
When the user attempts to withdraw $20
Then the transaction should fail
And the account should have a balance of $20

Given a user has a bank accout
And the account has a balance of $10
When the user attempts to withdraw $5
Then the transaction should succeed
And the account should have a balance of $5

Given a user has a bank accout
And the account has a balance of $10
When the user attempts to deposit $5
Then the transaction should succeed
And the account should have a balance of $15

Doing something like this is more elegant. The steps definitions would be

• a user has a bank account performs setup tasks to load a test user into the system.
• the account has a balance of X would either load the account with that much or just make an initial deposit.
• the user attempts to withdraw X/the user attempts to deposit X would both perform those actions. If it’s a web API they’d be HTTP requests.
• then the transaction should X would check HTTP status code. You could make this two separate step definitions or have logic based on the word. Two separate step definitions is probably better.
• the account should have a balance of X checks the balance in the account.