If you want to reproduce the type of additive blending you'd see in a photo editing app like Photoshop, you probably want to have both the source and the destination in a gamma-corrected color space such as sRGB. If you want to reproduce how light actually works, you should leave your colors in linear RGB.
When photo editing apps first implemented blend modes, they were done in a gamma-corrected color space. Doing an add of 2 colors in such a space (especially in 8-bits per channel) would cause a lot of clipping of colors. Because of this, there were blend modes that achieved some of the desired light-like effects in the gamma-corrected space. For example, the "screen" blend mode in gamma-corrected space looks a lot like an "add" blend mode in linear space.
Fast-forward 20 years and these apps started adding linear RGB as an option for editing and suddenly all the blend modes look "weird" (at least to users who had been doing gamma-corrected blending for the last 20 years). So some applications do a gamma correction before doing blending, and then convert the result back to linear afterwards. So if that's working for you, it's an absolutely fine way to do it. You wouldn't be the first!
See for example, this book:
Artists have dealt with the problems of working directly in video space [i.e. gamma-corrected space] for years without even knowing. A perfect example is the Screen transfer mode, which is additive in nature but whose calculations are clearly convoluted when compared with the pure Add transfer mode. Screen uses a multiply-toward-white function with the advantage of avoiding the clipping associated with Add. But Add's reputation comes from its application in bright video-space images. Screen was invented only to help people be productive when working in video space, without overbrights; Screen darkens overbrights (Figures 11.11a, b, and c). Real light doesn't Screen, it Adds. Add is the new Screen, Multiply is the new Hard Light, and many other Blending modes become evident as mere kludges in linear floating point
(a^gamma+b^gamma)^(1/gamma)
is not equal to(a+b)^(1/gamma)
\$\endgroup\$