Formal Specification
Capturing Bases / Base-to-Base Tag — Formal Specification
Philippines · chase-tag · Confidence: low
← Back to game pageImplementation Suitability
Suitability: engine-readyFidelity: discretizedGap-fills: 5
# Agawan Base — AI-optimized specification
## General idea
A simultaneous-tick 2D continuous/grid multi-agent team strategy game where two opposing squads defend home base landmarks located at and . Tagging priority between opposing field agents is determined strictly by the temporal recency of departure from their respective home base (the agent with the higher departure tick timestamp possesses tagging rights). Tagged agents are immobilized at the opponent's base in a rescueable chain; touching the opponent's base or capturing all enemies scores a game victory.
## Formal model
- **Abstraction taken:** Arena is discretized into a 2D integer grid of size (1-meter resolution). Base A is positioned at ; Base B is positioned at . Agents touching their home base coordinate have `immune = true` and update their `last_depart_tick = current_tick` upon leaving the base cell.
- **Player/agent count:** 2 teams of 5 agents each ( and ).
- **Turn model:** Simultaneous discrete execution ticks ( step per round).
## Formal rules
R1. Setup:
- Team A agents spawn at Base A with `state = AT_BASE`, `last_depart_tick = 0`.
- Team B agents spawn at Base B with `state = AT_BASE`, `last_depart_tick = 0`.
R2. Movement & Seniority Timestamp Updates:
- Every tick, each active non-prisoner agent submits a movement vector .
- If an agent moves off their home base cell, `state` becomes `IN_FIELD` and `last_depart_tick` is locked to `current_tick`.
- If an agent in field returns to their home base cell, `state` becomes `AT_BASE`, and `last_depart_tick` is refreshed on next exit.
R3. Tagging Resolution:
- If agent (in field) and agent (in field) are at distance unit:
- If : is tagged (`state = PRISONER_AT_A`).
- If : is tagged (`state = PRISONER_AT_B`).
- If : Mutual clash; both agents are repelled 2 cells back toward their base without capture [gap-fill].
R4. Prisoner Chain & Rescue:
- Prisoners at Base A form a chain: Prisoner 0 is at ; Prisoner is placed at [gap-fill].
- If an active free teammate reaches distance of any member of the prisoner chain without being tagged, all chained prisoners transition to `state = RETURNING_FREE` (immune from tagging for 15 ticks while returning to Base B) [gap-fill].
R5. Win Conditions:
- Base Infiltration: If an active agent reaches Base B cell , Team A immediately wins the round (+10 points).
- Total Capture: If all 5 agents of a team become prisoners, the opposing team wins (+10 points).
- Time Horizon: If no base capture occurs within ticks, the team with fewer captured prisoners receives +5 points [gap-fill]. Match concludes after 3 rounds.
## Edge cases
- Tie at the end condition: If match points are tied after 3 rounds, a 1v1 sudden death sprint on a reduced 20x10 field is played [gap-fill].
- A player with no legal move: (staying at base or field) is always legal.
- Repetition / cycles: Bounded by the hard countdown limit ticks per round.
- Simultaneous claims on the same resource or position: When multiple agents enter the base coordinate simultaneously, tagging priority resolves first; if an un-tagged attacker enters, base capture triggers immediately.
- Player elimination mid-round: Captured agents become static chain nodes until rescued, preserving exact agent array bounds.
- Resource exhaustion: Not applicable.
- Termination guarantee: Enforced by the finite tick ceiling .
## State representation
- Persistent state:
- `agent_positions`: array of 10 coordinates .
- `agent_states`: array of 10 enums (`AT_BASE`, `IN_FIELD`, `PRISONER_AT_A`, `PRISONER_AT_B`, `RETURNING_FREE`).
- `depart_timestamps`: array of 10 integers.
- `immunity_timers`: array of 10 integers.
- `team_scores`: array of size 2.
- `current_tick`: integer .
- Observable information: Fully observable public state (all agent positions, base states, and timestamps are common knowledge).
## Action space
- `move(dx, dy)`: with .
## Playability notes
- Single-player-vs-AI: Highly strategic team simulation; a player controls a squad leader coordinating baiting and base-rush tactics.
- Group games (n>4): Reduces smoothly to 2v2 or 3v3 on a 25x15 grid without altering the temporal seniority mechanic.
- Physical-skill games: Sprint speed is parameterized by movement steps per tick; tactical timing and spatial coordination are preserved intact.
- What is lost in the adaptation: Physical diving reaches, deceptive body momentum, and chain-link hand grip strength.
## Implementation hints
Discrete 2D vector field with timestamp comparison logic. Branching factor is joint moves per team per tick. First milestone: Implement 2v2 simulation with fixed straight-line paths to validate timestamp priority and prisoner rescue chain mechanics.
Download as JSON
Download the full game corpus entry for this game as structured JSON.
Download JSONCite this page
Capturing Bases / Base-to-Base Tag (Philippines). (n.d.). Games of Countries. https://gamesofcountries.com/games/ph-philippines/agawan-base
@misc{gamesofcountries-ph-philippines-agawan-base,
title = {Capturing Bases / Base-to-Base Tag (Philippines)},
year = {n.d.},
publisher = {Games of Countries},
url = {https://gamesofcountries.com/games/ph-philippines/agawan-base}
}