NewJerseyStyle/plugin-mcts Skills

Execute Monte Carlo Tree Search (MCTS-LLM) to solve complex problems through iterative exploration, simulation, and learning. Use this for research questions, planning tasks, and coding challenges that benefit from systematic exploration of solution spaces.

Execute the BACKPROPAGATION phase of MCTS to update node statistics from leaf to root

Manage the MCTS prompt dataset with CRUD operations, import/export, and reset functionality. Use for viewing, editing, backing up, and restoring prompt templates.

Execute the EXPANSION phase of MCTS to generate new child nodes using LLM as world model

Execute the SELECTION phase of MCTS using UCB1 to traverse from root to a promising leaf node

Execute the SIMULATION (rollout) phase of MCTS using LLM as heuristic policy to evaluate a node