The typical tabula recta is rather indistinct, visually, so I took a few tries at making it easier to use.
With more subtle delineation, the eye can more confidently trace from the plaintext letter, to the key letter, to the cipher letter.
This first attempt at delineation demonstrates how strong contrast isn't good either, is less easy on the eyes, and thus, not the ideal choice.
This is a tabula recta. It's a table of the alphabet with each subsequent row shifted one to the left. The letter that is bumped off is added onto the end of that row. It's for cryptography, specifically, the polyalphabetic Vigenère cipher, aka le chiffre indéchiffrable (French for 'the indecipherable cipher'), wherein plaintext is encoded via a repeated keyword and the following table. There is a another method, called the Variant Beaufort, which is a reciprocal cipher, meaning the plaintext is encrypted via the decryption method, and vice versa... but that is merely one step beyond and easily checked for as well.
I'm pretty sure that nobody except cryptography nerds use a tabula recta anymore. There are sites and scripts online to decrypt ciphertext with known keywords/phrases. However, it was generally invincible for about 300 years until Charles Babbage, and later, more usefully, Friedrich Kasiski, discovered that it was susceptible to frequency analysis just like any monoalphabetic ciphertext would be, if only extrapolated to guess and check for key length.
Say I want to encode, "my spoon is too big." with the keyword, "pickle". For the letter, "m", I would look at the tabula recta and find, "m", then look right, to the letter, "p", then up to the top and use that letter, "d", in the ciphertext. You can extrapolate the rest from that simple method. At the end of the keyword, just repeat: