IDML is a deep format and the corpus never runs out of edge cases. Issues, sample documents, and parser fixes are all contributions — the engine repo is public, the conformance harness runs anywhere.