Skip to content

Latest commit

 

History

History
68 lines (49 loc) · 2.84 KB

File metadata and controls

68 lines (49 loc) · 2.84 KB

INTEL.md -- Cross-Game Intelligence

This is the top-level accumulator. It contains only distilled insights that apply across multiple games. Nothing gets promoted here until a second game confirms the pattern.

Intelligence flows up:

Scenario INTEL  →  Game INTEL  →  This file
(per-battle)       (per-game)     (universal, 2+ games required)

Update protocol:

  • After every game onboarding: add a Game Log entry (Section 1 only)
  • Patterns stay in the game-level INTEL until a second game confirms them
  • Once confirmed by 2+ games: promote to Sections 2-4 with provenance tags
  • Never store game-specific details here

1. Game Log

Registry of onboarded games. This is an index only -- game-specific intelligence lives in games/<GameName>/INTEL.md.

# Game System Family Onboarded Game INTEL
1 SPQR Great Battles of History (GBoH) 2026-04-10 games/SPQR/INTEL.md

2. Mechanical Patterns

Cross-game insights confirmed by 2+ games. (Empty until a second game is onboarded.)

Candidates awaiting second confirmation (from single-game INTEL files):

  • Directional ZOC creates flanking opportunities
  • Cohesion systems demand proactive withdrawal
  • Activation systems reward sector focus
  • Multi-board map modules need active board detection from save (BoardPicker setup commands)
  • Hex grid parameters MUST be extracted from buildFile.xml, never guessed
  • VASSAL state strings use nested SequenceEncoder escaping (each Decorator level adds backslashes)
  • Leader Active/Finished state is in BasicPiece name suffix (/1 vs /2)
  • Cohesion Hits / damage levels stored as Embellishment levels in OUTER->INNER chain order

3. AI Strategy Principles

Universal strategic insights confirmed by 2+ games. (Empty until a second game is onboarded.)

Candidates awaiting second confirmation:

  • Calculate the rout/break threshold before committing to combat
  • Cavalry with pursuit rules is a strategic commitment, not a tactical one
  • Leader preservation is paramount in leader-dependent activation systems
  • Screen with expendable light units
  • Monte Carlo with N=300-500 iterations is sufficient for tactical decisions
  • Adjacency + ZOC + matchup superiority are the 3 critical factors for tactical wargames
  • AI should prefer FREE units for movement; LOCKED units for shock combat
  • Asymmetric matchups (heavy vs light): heavy almost never voluntarily attacked by light; ZOC stalemates favor whoever can't be hurt

4. Module Pattern Recognition

Patterns in how VASSAL modules encode game data, confirmed by 2+ modules. (Empty until a second module is analyzed.)

Candidates awaiting second confirmation:

  • DynamicProperty with short names for primary damage tracking
  • Terse prototype names mapping to unit categories
  • d10 configured as 1d10-1 for 0-9 results