Simple Agent
This agent expects an “OPENAI_API_KEY” Environment Variable to exist in the system, and to contain contain a valid openAI key.
Core Functionality
This agent follows within the following workflow:
- Chat Handling
- Receives player messages
- Analyzes conversation context and game state
- Generates relevant responses about the current puzzle
- Move Generation
- Evaluates the current board state via images
- Considers the objective shape and conversation history
- Decides on a move and sends it using gpt4o
- Feedback Processing
- Receives information about move validity
- Makes adjustments based on collision detection
- Iteratively reviews the result with gpt4o and sends adjusted plays until the shape is properly placed.
- Limits adjustment attempts to prevent endless loops
- The agent can make up to 10 adjustment attempts before forcing a finish