@mindme showed how Tito uses influence maps to spread creep across the map, hitting 80% coverage by 10 minutes with just two queens. @chad demo’d Test Lab, a local automated testing framework that runs your bot against custom bots, past versions, and even itself, with git hook integration for commit-triggered test suites. We also touched on the Ralph Loop workflow for autonomous bot coding with validation.
Key Takeaways:
Influence maps can drive creep spread by weighting positions toward the enemy natural and base-to-base paths, keeping tumors spread out instead of clumped
Add negative weights around detectors and enemy zones so your creep avoids getting killed before it reaches useful positions
Automated testing against in-game AI and downloaded ladder bots gives you a fast feedback loop on whether your changes actually helped
Git hooks that trigger test suites on commit let you catch regressions before you push, no manual replay watching needed
Mirror matches (bot vs itself) are a cheap way to surface weaknesses, one side always loses and that loss is your next fix