The first thing to note is that if your goal is to import data into a lot of similar graphs, usually the best option is to first save out a blank template file with your preferred styles, and then use OmniGraphSketcher to copy/paste the data into duplicates of the template. However, I do realize there are limitations with this approach, in particular that it is not amenable to setting up batch jobs for creating large numbers of graphs.
So... the next option is making your own XML files. This comes with the usual disclaimer that the XML format could change in future versions of OmniGraphSketcher and "your mileage may vary." But we did choose an XML file format so that you would have this option.
Your best bet for a "template" is still going to be a blank graph file with your preferred styles applied. Save the template, change the extension to .zip, unzip and find the contents.xml inside the zip archive. Copy that XML file into a new location and rename it to have a .ograph extension; OmniGraphSketcher will be able to read it directly (saving you zipping/unzipping hassles as you experiment).
The schema for our XML format is documented in detail as RelaxNG. (See relaxng.org
for more info about RNG and how to validate your XML files.) To find the .ograph schema, right-click on the OmniGraphSketcher app, choose "Show Package Contents", then navigate to Contents -> Resources -> GraphSketcher.rng . Almost all attributes have defaults, and most of these are documented in the RelaxNG file.
What you will most likely want to do is leave the entire template xml file intact, and insert your own graph elements at the end, just before:
The main limitation you will probably run into is that automatic axis scaling doesn't work. That is, the axis ranges and some of the tick labels have to be specified exactly in the XML file. Your best bet is probably to leave the axis ranges and tick labels as-is in your sample file. Then, after finishing your XML processing, you can open the resulting file in OmniGraphSketcher and just choose Arrange -> Scale to Fit Data.
Hope this helps! Let me know if anything needs clarification.