Ditch the image

use this format for board representation

``` { "game_state": { "board_size": { "width": 9, "height": 9 }, "mines_total": 10, "mines_flagged": 2, "game_status": "in_progress", // "in_progress", "won", "lost" "time_elapsed": 45, "difficulty": "beginner" // "beginner", "intermediate", "expert", "custom" }, "board": [ ["1", "?", "?", "2", "1", "1", "1", "1", "0"], ["1", "2", "?", "2", "?", "1", "1", "?", "0"], ["0", "1", "1", "2", "1", "1", "1", "1", "0"], ["0", "0", "0", "0", "0", "0", "0", "0", "0"], ["1", "1", "0", "0", "0", "0", "0", "0", "0"], ["?", "1", "0", "0", "0", "1", "1", "1", "0"], ["1", "1", "0", "0", "0", "1", "F", "1", "0"], ["0", "0", "0", "0", "0", "1", "1", "1", "0"], ["0", "0", "0", "0", "0", "0", "0", "0", "0"] ], "last_action": { "action_type": "reveal", "x": 3, "y": 2, "result": "revealed_number", "timestamp": 1710931245 } } ```

and this format for llm response generation

``` { "action": { "action_type": "reveal", // "reveal", "flag", "unflag", "chord" "x": 5, "y": 3, "confidence": 0.95, "reasoning": "This cell is surrounded by revealed cells with low numbers, making it a safe choice." }, "game_analysis": { "identified_safe_cells": [[5, 3], [2, 5]], "identified_mine_cells": [[6, 1], [8, 2]], "uncertain_cells": [[1, 1], [2, 2]], "strategy": "Targeting isolated revealed areas first to gain more information." } } ```

It should fix all your issues plus also make it cheaper to play