Should I extend lammps in the way I'm going to?

Dear Lammps developers,
I have a need of computing rdf, centered at points of domain that have enough free space around the (in some measure). To use “compute rdf” command, I need a way of defining those center points, so I’m planning to extend lammps with a custom fix style, which will compute those points as atoms of a special style for further passing it to “compute rdf”. In more details, the custom fix command will perform discretizations of the domain, searching for clusters of adjacent free cells and computing their mass centers.
I have not decided anything about the name or arguments yet, just writing to ensure that my design is acceptable.
Thank you in advance!

Dear Lammps developers,
I have a need of computing rdf, centered at points of domain that have
enough free space around the (in some measure). To use "compute rdf"
command, I need a way of defining those center points, so I'm planning to
extend lammps with a custom fix style, which will compute those points as
atoms of a special style for further passing it to "compute rdf". In more
details, the custom fix command will perform discretizations of the domain,
searching for clusters of adjacent free cells and computing their mass
centers.
I have not decided anything about the name or arguments yet, just writing
to ensure that my design is acceptable.

​a couple of comments.

- your strategy​ is similar to what pair_style srp does (and the
corresponding fix SRP, which is generated by pair srp), so you may want to
study those codes.
  pair srp adds a repulsive spherical interaction at the center of (long)
bonds, so that bonds cannot cross.

- if you add virtual particles, you have to be careful to correct the
degrees of freedom of the system accordingly, or else your thermodynamic
properties may be wrong and you need to manage neighbor list exclusions for
efficiency.

​axel.​