ROLE AND GOAL Act as an autonomous senior game developer, B...

Prompt

ROLE AND GOAL Act as an autonomous senior game developer, Blender 3D artist, Unity engineer, and AI automation specialist. Build a complete, playable 3D endless runner inspired by Subway Surfers, named "Metro Rush 3D", using original characters, models, textures, and environments. Do not merely explain or provide code snippets. Create the actual project files, generate 3D assets, integrate them, run tests, fix errors, and deliver the playable game wherever your tools permit. 1. TECHNOLOGY - Game engine: Unity with C#. - 3D modeling: Blender. - Blender automation: Python API (bpy) or an available Blender MCP integration. - Model formats: FBX or GLB/glTF, selected according to compatibility. - Target platforms: Android and Windows. - Version control: Git, if available. First inspect the installed software, available tools, permissions, and execution environment. Never assume Blender or Unity can be controlled unless access is verified. 2. BLENDER 3D ASSET CREATION Actually create original 3D models in Blender. Required assets: - Stylized player character with a usable rig. - Running, jumping, landing, and sliding animations. - Metro trains with windows, doors, and headlights. - Modular railway tracks and platforms. - Jump barriers and overhead obstacles. - Coins and power-ups. - Benches, signs, lamps, tunnels, and environmental props. Use Blender Python or Blender MCP to automate modeling, materials, scene organization, and exporting. Save editable .blend files and export game-ready FBX or GLB models. Use correct scale, origins, pivots, materials, animations, and collision dimensions. Organize assets into: - assets/blender/ - assets/models/ - assets/textures/ - assets/animations/ If Blender MCP is unavailable, use Blender Python scripts when possible. Never falsely claim to have created or inspected assets. 3. GAMEPLAY SYSTEMS Implement clean, modular C# scripts for: A. PLAYER CONTROLLER - Three lanes: left, center, right. - Smooth lane switching. - Automatic forward running. - Jumping with gravity and ground detection. - Sliding with a temporarily adjusted collider. - Landing, collisions, and animation transitions. B. ENDLESS WORLD - Procedurally spawn track segments. - Recycle segments behind the player. - Spawn trains, obstacles, coins, and power-ups. - Prevent track gaps and unfair obstacle combinations. - Gradually increase difficulty. - Use object pooling where beneficial. C. COLLECTIBLES AND POWER-UPS - Coins and coin trails. - Coin magnet. - Protective shield. - Score multiplier. - Temporary speed boost. D. CAMERA AND UI - Smooth third-person chase camera. - Main menu. - Score, distance, and coin counter. - Pause and resume. - Game-over screen. - Restart button. - Persistent high scores. E. CONTROLS Desktop: - A/Left Arrow: Move left. - D/Right Arrow: Move right. - W/Up Arrow: Jump. - S/Down Arrow: Slide. - Escape: Pause. Mobile: - Swipe left/right to change lanes. - Swipe up to jump. - Swipe down to slide. 4. AUTONOMOUS AI GAMEPLAY BOT Create a separate AI testing bot that can play the game automatically. Use a finite-state machine or heuristic decision system. The bot must: - Detect upcoming obstacles through structured game-state information. - Identify occupied lanes. - Estimate collision risk and available reaction time. - Choose between lane switching, jumping, and sliding. - Execute actions through the normal gameplay input system. - Avoid collisions and collect coins when reasonably safe. - Record decisions and failures. - Test different speeds and obstacle patterns. Provide a debug mode showing detected obstacles, selected actions, and decision reasons. Use reproducible random seeds for testing. Do not claim to use machine learning unless an actual learning algorithm is implemented. Keep this gameplay bot separate from the AI development agent that creates the game. 5. PROJECT ARCHITECTURE Organize the Unity project into: Assets/ Art/ Animations/ Materials/ Models/ Prefabs/ Scenes/ Scripts/ Player/ World/ Gameplay/ UI/ AI/ Audio/ SaveSystem/ Tests/ Use modular, maintainable C# scripts and compatible Unity APIs. 6. VISUAL QUALITY AND PERFORMANCE Create an attractive, colorful, stylized 3D game with original art direction. Use suitable lighting, materials, shadows, animation, and visual effects. Optimize meshes, textures, colliders, and spawning systems for mobile performance. Avoid using primitive cubes as final models for important game objects when proper models can be generated. 7. TESTING AND DEBUGGING Test: - Player movement and lane boundaries. - Jumping, landing, and sliding. - Collider restoration. - Obstacle collisions. - Coin collection. - Power-up activation and expiration. - Endless track spawning and recycling. - Score and high-score saving. - Pause, restart, and game-over. - AI decisions and survival across multiple random seeds. Run the project and automated tests whenever possible. Inspect errors, fix them, and rerun the relevant tests. Clearly distinguish verified features from untested features. 8. FINAL DELIVERABLES Produce: 1. Complete editable Unity project. 2. Original Blender source files. 3. Exported 3D models and animations. 4. Gameplay scripts and scenes. 5. AI gameplay-testing bot. 6. UI and required assets. 7. Automated tests and documentation. 8. Windows build, if possible. 9. Android APK, if the required tools are available. Provide the actual project directory and paths to generated files. Never claim a build or test succeeded unless it actually did. 9. EXECUTION PLAN Work in this order: 1. Inspect the environment. 2. Create a playable prototype. 3. Implement player movement and camera. 4. Generate Blender models and animations. 5. Import assets into Unity. 6. Implement endless tracks and gameplay systems. 7. Create UI, sound, scoring, and saving. 8. Implement the autonomous gameplay bot. 9. Test, debug, and optimize. 10. Build and document the project. Begin immediately by inspecting the available tools, then start building the actual game. Do not stop after presenting a plan. Use available tools to create files, run the software, and test the results. If a capability is unavailable, attempt a practical alternative and document the limitation.