Repairing flawed domain models remains a critical challenge in AI planning, with few effective techniques available. We propose a novel approach for repairing totally ordered hierarchical task network (TO-HTN) models with missing actions, guided by a plan that must be valid for the repaired model. This problem has only one previously documented approach, which relies on complex re-encoding that's solved via TO-HTN planning. In contrast, our approach translates the repair task into a context-free grammar repair problem and leverages a large language model (LLM) to identify and insert relevant actions directly, simplifying the repair process. We evaluate our approach on established benchmarks and demonstrate substantially improved results over the prior approach, achieving nearly three times the number of instances solved, and nearly solving all instances of domains in which the previous approach solved zero. Importantly, we mask all natural language hints, such as action names, forcing the LLM to simulate reasoning and planning, and mitigating the risk of data leakage from its training corpus.

AI agents have successfully employed deep reinforcement learning methods to surpass human performance in various new tasks over the past decade, notably including the domain of games. However, Angry Birds requires complex physical and spacial reasoning that is yet to be captured by such means. We present our logic-based Angry Birds AI which won the 2021 IJCAI AIBIRDS competition and propose a simple new method we call Second Order Thompson Sampling (SOTS) which allows for fine-tuning the balance between exploration and exploitation. We cover the competition scores of our entrant Agent X, its predecessor - the former state of the art Bambirds 2019, and the new and improved Bambirds 2021. We find that our agent has the best all-round performance, but would gain a lot by incorporating the improvements of Bambirds 2021. We list other potential areas of improvement for a future superhuman Angry Birds AI.