ai-code-battle/notes/bf-3rr.md

Leader-Targeter Bot Implementation (Bead bf-3rr)

Summary

The leader-targeter bot was successfully implemented in Java with Maven build system.

Implementation Details

Strategy

The bot implements multi-player game theory by:

1. Score Calculation: Estimates opponent scores based on visible bots (10 points each) and active cores (2 points each)
2. Target Selection: Identifies the score leader and directs all units toward them
3. Tiebreaking: Uses nearest distance when scores are tied
4. Core Defense: Detaches 2 bots to defend own core when enemy is within 6 tiles
5. 2-Player Fallback: Falls back to straight aggression when only one opponent exists

Files

• bots/leader-targeter/src/main/java/com/acb/targeter/App.java - HTTP server with signature verification
• bots/leader-targeter/src/main/java/com/acb/targeter/LeaderTargeterStrategy.java - Core strategy implementation
• bots/leader-targeter/src/main/java/com/acb/targeter/GameState.java - Game state parsing
• bots/leader-targeter/Dockerfile - Docker build configuration
• bots/leader-targeter/pom.xml - Maven build configuration

Build

• Maven shade plugin creates executable JAR with bundled dependencies
• Javalin HTTP server on port 8085
• Jackson for JSON processing
• Java 21 target

Retrospective

What worked

The Maven project structure with Javalin and Jackson provides a clean foundation for bot implementations. The strategy correctly models the N-player problem and applies the kingmaker dynamic.

What didn't

N/A - implementation was straightforward

Surprise

The bot was already fully implemented in a previous commit, so this task primarily involved verifying the build and pushing existing work.

Reusable pattern

• Maven shade plugin configuration for executable JARs
• Toroidal distance calculations for grid-based games
• Centroid calculation for targeting multiple assets