Skip to main content
Formal Specification

Fortress Tag / Base Capture — Formal Specification

Indonesia · chase-tag · Confidence: low

← Back to game page

Implementation Suitability

Suitability: engine-readyFidelity: discretizedGap-fills: 3
# Bentengan — AI-optimized specification ## General idea A multi-agent territory and tagging game where two teams contest an open 2D space between two fortress nodes. The central mechanic is temporal priority (*freshness hierarchy*): an agent who left their home base more recently can capture any enemy agent who departed earlier. Agents balance baiting, prisoner rescue, base defense, and terminal fortress infiltrations. ## Formal model - **Abstraction taken:** Continuous physical field is represented as a 2D integer grid [0..40]×[0..20]. Base 0 is located at (0,10) and Base 1 at (40,10). - **Player/agent count:** 2 teams of N agents (N∈[2..6], default 4 per team). - **Turn model:** Simultaneous tick execution (Δt=1 step). ## Formal rules R1. State setup: Base 0 at (0,10), Base 1 at (40,10). Agents of Team 0 start at Base 0 with timestamp τ=0. Agents of Team 1 start at Base 1 with timestamp τ=0. R2. Movement & base refresh: Each tick, every non-captured agent chooses a move v→∈{(dx,dy):|dx|≤1,|dy|≤1}. - If agent i of Team k occupies base position Basek, their timestamp updates to the current tick: τi=t. R3. Tagging & capture: If enemy agents A (Team 0) and B (Team 1) are within Euclidean distance d(A,B)≤1.5: - If A is on Base 0, A is safe; B cannot tag A. - If B is on Base 1, B is safe; A cannot tag B. - In open field: If τA>τB, A captures B. B becomes a prisoner (`status = PRISONER`) and is moved to prisoner node (0,12) near Base 0. - If τB>τA, B captures A. A becomes a prisoner (`status = PRISONER`) and is moved to prisoner node (40,12) near Base 1. - If τA==τB, neither captures the other (neutral bounce) [gap-fill]. R4. Rescue: If an active agent C of Team 0 reaches within distance d(C,P)≤1.5 of prisoner P (teammate), P is immediately liberated (`status = ACTIVE`, τP=t). Both C and P receive 10-tick invulnerability to return to base [gap-fill]. R5. Win condition: - If any active agent of Team 0 reaches distance ≤1.0 of Base 1 without being tagged, Team 0 wins immediately (Base Sieged). - If any active agent of Team 1 reaches distance ≤1.0 of Base 0 without being tagged, Team 1 wins immediately. - If all agents of a team are simultaneously prisoners, the opposing team wins. - [gap-fill] If neither team achieves a win condition by tick Tmax⁡=1000, the team with fewer prisoners (or closer net field position) is awarded the win. ## Edge cases - Tie at the end condition: If both bases are touched on the exact same tick, or at Tmax⁡ with identical metrics, the game is declared a draw. - A player with no legal move: (0,0) (remain stationary) is always legal. - Repetition / cycles: Bounded by max tick limit Tmax⁡=1000. - Simultaneous claims on the same resource or position: Grid cells can be multi-occupied; tagging is resolved pair-wise using timestamp inequality τA≠τB. - Player elimination mid-round: Prisoners remain on field at detention coordinates and can be dynamically rescued; turn order and tick progression continue without interruption. - Resource exhaustion: Not applicable. - Termination guarantee: Enforced by the finite tick horizon Tmax⁡. ## State representation - Persistent state: - `agents`: array of objects `{id, team, x, y, timestamp, status, invulnerable_ticks}`. - `score`: `{team_0, team_1}`. - `current_tick`: integer 0..1000. - Observable information: Fully observable (all positions, statuses, and timestamps are public knowledge). ## Action space - `move(dx, dy)`: dx,dy∈{−1,0,1}. ## Playability notes - Single-player-vs-AI: Very playable as a tactical squad management or 1-agent control game with rule-based/RL bots. - Group games (n>4): Naturally reduces to 2-agent or 3-agent squads per side without losing core timestamp bait-and-switch dynamics. - Physical-skill games: Physical sprinting and dodging speed are normalized into discrete directional vectors; the decision tree focuses on timestamp management and spatial coordination. - What is lost in the adaptation: The sudden physical agility of dodging arm-reach tags at full sprint is abstracted into a distance check. ## Implementation hints Extremely engaging multi-agent environment with emergent strategies (baiting high-timestamp enemies away from base while a refreshed flanker charges the fortress). First milestone: Implement 2v2 on a 20×10 grid with timestamp tracking and base capture win triggers.

Download as JSON

Download the full game corpus entry for this game as structured JSON.

Download JSON
Cite this page

Fortress Tag / Base Capture (Indonesia). (n.d.). Games of Countries. https://gamesofcountries.com/games/id-indonesia/bentengan

@misc{gamesofcountries-id-indonesia-bentengan,
  title = {Fortress Tag / Base Capture (Indonesia)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/id-indonesia/bentengan}
}