Two suggestions for projection system authors.
I'm a big fan of baseball projections. They're a welcome sight around the in the doldrums between seasons, and they're fun to debate.
In prototyping a unified dashboard for mixing and remixing projections (more on that later), I've spent ample time with most major distributions. There are two points, I feel, that would make working with this data less troublesome.
Include a license.
Including a license is an essential step for anyone releasing data into the wild. Without a license, it's terminally unclear if I can build around your data; knowing what rights and attribution you retain in your distribution would be great. At the very least, include note about how I may use your data. And, explicitly stating “no license” tells me more than not having a license ever could.
Strive to include at least one globally-available player ID.
Not all player names are unique. If you're projecting major and minor leaguers, you're going to have name duplication.
The Chadwick Bureau has done the hard work to build a player Rosetta Stone incorporating Retrosheet, MLBAM, FanGraphs, B.Pro, NPB, and more. It's not perfect, but it should prove helpful in filling gaps. It has for me.
I realize this potentially introduces more complexity than authors can afford or can afford to care about. Failing that, please include as many peripherals (team, age, bats, throws) as possible. Matching using only these data points is not ideal, but their availability goes a long way toward disambiguation. These peripherals are also what could be used in the case where a player does not have a database from any major source.
Split player names into first and last names.
First and last names are always useful, especially when disambiguating without an ID. Please consider splitting names if possible.