Introductory programming courses face the challenge of providing high-quality, personalized feedback to a diverse student population. Our research aims to improve automated feedback by bridging the gap between complex industrial-strength static analysis tools and the practical needs of non-expert teachers. To achieve this, we are developing Pyttern, an accessible Program Query Language (PQL) based on the code-with-holes paradigm. This research statement presents our progress in formalizing Pyttern’s matching semantics using Tree Pattern Automata (TPA), extending it with sub-pattern calls for modularity, and outlining future plans for mining structural code patterns or using Pyttern for more advanced use cases such as design pattern detection.