Observation: what game state can the agent read? Record whether it receives pixels, structured state or both.
Actions: what commands can it issue? Distinguish model decisions from helper code that performs movement or pathfinding.
Environment: identify the server or test setup, starting conditions and any interventions.
Result: note the actual goal, time, failures and completion evidence. Read limitations alongside highlights.
02
Your experiment notes
Agent/model: __________
Observations and available tools: __________
Test environment and starting state: __________
Human interventions or helper systems: __________
Goal completed, evidence and failure cases: __________
03
Design your own small test
Use a controlled environment where automation is permitted. Choose one measurable task and keep the setup reproducible.
Record failures as well as successes and change only one condition between comparisons.
FROM THE ORIGINAL CREATOR
This companion guide is by Huzaifa. Check the original project for current setup and availability.
