AI Aggression recommendation

Maybe hard-code some values by discipline. Maybe crowd-source.

The issue mentions AI aggression recommendations by discipline, but gives no specifics on where in the codebase this logic resides. Start by exploring the repository structure for AI-related files, possibly under src/ai or similar directories. Look for existing aggression settings or discipline definitions. Determine how values could be hard-coded or crowd-sourced, and what the expected output format is.

索引モデルが issue の本文から書いたものです。