Some people have a knack for creating tsumego. Is there anyway to develop this knack? Looking at various accounts on Goproblems.com it seems some are better than others at making tsumego, though I suspect many are copies, I just did one that is in my Lee Changho tesuji book for example. Also it seems like it would be a useful defensive skill take for example Cho U's expertise.

Practice. How many things are you actually good at the first time you do them?

_________________
We don't know who we are; we don't know where we are.
Each of us woke up one moment and here we were in the darkness.
We're nameless things with no memory; no knowledge of what went before,
No understanding of what is now, no knowledge of what will be.

Good point, however, in many endeavors, I am better the first time if I do a little research prior.

Just taking them from a game or modifying an existing tsumego can be a quick approach to it. But if you want to write them from scratch, here's one approach:

Start with the solution (say, a snapback, or using a monkey jump to connect, or creating a double atari...) Put the basic stones needed for the tesuji to apply, then work backwards a little so that the tesuji brings life or death. You've now got an obvious, extremely simple tsumego. So you want to make it more difficult:

Tsumego difficulty is largely a factor of 3 things: How unusual is the right move (e.g., solutions involving empty triangles)? How many reasonable choices do you have? How many moves away is the solution?

An unusual right move is largely a matter of selecting an appropriate tesuji. It may also be achievable by adding extra stones (e.g., if the correct move is a solid extension, can you add a stone so that it's now an empty triangle?)

The number of reasonable choices can sometimes be increased with extraneous stones. Add opportunities to almost, but not quite, form an eye. Make opposing groups have just enough liberties that they can be threatened but win any capturing race. Sometimes you'll have too many stones and need to remove them. Read that the solution is the same without a stone, and see if removing it makes the position look more 'open'. Leaving cutting opportunities in the player's position can help, forcing them to read that the cut doesn't work.

Finally, there's making the solution require more moves. There's a number of ways to do this: create two problems that each produce (or poke out) a single eye and join them. Make sure there's a way to get to play both of them. Or remove key stones from the solution, and then extend the problem so that they can be placed down again as forcing moves (e.g., make a move miai for forming a simple eye, or executing the solution tesuji). Weak opposing groups can be a good way to chain tesuji together: E.g., you can escape by capturing a white group. But it stays one move ahead until escaping to safety. That chase, though, reduced the liberties of a second group. So now a hane forces it to connect out. And that hane stone is the last needed to execute a snapback in the corner...

Just iterate between changing each of those until you like the problem. And then read very carefully that the solution you think works, works, and that there aren't any easier solutions.