Cracking an XOR crypt with a know key length

I'm trying to crack a crypt with a known key length. I deduced that the operation made was a hex XOR. Here is the crypt:

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

I have tried to use this tool to decrypt it. The tool outputted multiple possible keys and for each key, an attempt do decipher the crypt. I know the result should be plain text English. The closest I got was this:

T-e po1ato ,s a 6tarc-y, t0bero0s cr*p fr*m th  per nnia) nig-tsha!e So)anumetube7osumeL. T-e wo7d po1ato (ay r fer 1o th  pla+t it6elf ,n ad!itio+ to 1he e!ibleetube7. Inethe ndesi whe7e th  spe&ies ,s in!igen*us, 1hereeare 6ome *thereclos ly r late! cul1ivat d po1ato 6peci s.P*tato s we7e in1rodu&ed o0tsid  theeAnde6 reg,on f*ur c ntur,es a"o,a+d ha3e be&ome $n in1egra) par1 of (uch *f th  wor)d's #ood 6uppl<. Iteis t-e wo7ld'sefour1h-la7gestefoodecropi fol)owin" mai?e, w-eat $nd r,ce.

After some digging and manually tweaking the text, I got this:

The potato is a starchy, tuberous crop from the perennial nightshade Solanum tuberosum L. The word "potato" may refer to the plant, itself, in addition to the edible tuber. In the Andes, where the species is indigenous, there are some other closely related cultivated potato species. Potatoes were introduced outside the Andes region four centuries ago, and have become an integral part of much of the world's food supply. It is the world's fourth-largest food crop, following maize, wheat and rice.

It unfortunately did not work. I am now trying to find a clue as to what I should be doing to find the answer.

Your problem is a very common one when starting to study techniques of cryptanalysis. Cryptanalysis relies in many cases heavily on statistical properties, and therefore rather than provide you with an accurate answer, a good cryptanalytic technique will instead reduce the solution space to a space that is easy to perform exhaustive search on.

To see how that applies to your issue, note that after correctly deducing that the cipher is a repeated-XOR cipher, and that the key size is 5 bytes long (probably via methods such as Hamming distance estimation) the next step is to fit the letter distribution of the underlying plaintext to the generalized distribution of letters in the (english) language.

But the problem is that this, in general, will not be a perfect fit because while the generalized letter frequency was calculated statistically over enormous amount of text, the plaintext at hand is quite short, and in general will be necessarily shorter than the text based on which the generalized frequency was established.

This is why both at this stage (and actually also at the stage where you estimate the key length) you should design the cryptanalytic code to take into account several of the most likely options, rather than assuming that the best fit is the correct one for the case at hand. For example, as you can see in the Wikipedia page I have linked, there are actually two generalized letter distributions to choose from: one based on texts, another one based on dictionaries. Those two can be your top two attempts for example, upon which you base the first two calculations of the possible key. It is also easy to construct other possible letter distributions by slightly tweaking the original one or consulting other sources. Note that even if you have to consult 1000 possible letter distributions before finding which one yields the correct key, this is still a big improvement over a brute force search over a 5 byte key: $2^{8\cdot5} = 2^{40}$ options is a lot more than $1000 \approx 2^{10}$. I hope that demonstrates the essential point here.