Formal Specification
The Handkerchief (Greek Handkerchief Snatch) — Formal Specification
Greece · chase-tag · Confidence: high
← Back to game pageImplementation Suitability
Suitability: engine-readyFidelity: discretizedGap-fills: 2
# Mantilaki — AI-optimized specification
## General idea
A discrete, round-based team pursuit duel played on a 1D linear coordinate space between two opposing team baselines and a central token. Matched pairs sprint to the center; an agent must decide whether to snatch the token and sprint back to base, or wait for the opponent to snatch and execute an immediate tag.
## Formal model
- **Abstraction taken:** The 1D field is modeled as discrete coordinates (default ).
- Team 0 Base: .
- Team 1 Base: .
- Handkerchief (Token): .
- **Player/agent count:** 2 teams (Team 0 and Team 1), each with agents. Each round activates 1 agent per team (e.g. and ).
- **Turn model:** Real-time ticked simultaneous execution per round duel.
## Formal rules
R1. Setup per duel:
- Referee calls number .
- Agent starts at , `has_token = False`.
- Agent starts at , `has_token = False`.
- Token sits at , `token_carried = False`.
R2. Simultaneous movement per tick :
- Each active agent chooses an action from:
- `sprint(forward)`: Move 2 units toward .
- `sprint(backward)`: Move 2 units toward own base.
- `creep(forward)` / `creep(backward)`: Move 1 unit.
- `hold()`: Stay at current coordinate.
- `snatch()`: Precondition: and `token_carried == False`.
R3. Snatch Resolution:
- If agent executes `snatch()` at :
- `A_i.has_token = True`, `token_carried = True`.
- From this tick onward, opponent is authorized to tag .
R4. Tagging & Foul Resolution:
- If (agents are adjacent or in the same cell):
- Case 1 (Legal Tag): One agent has `has_token == True`. Opponent tags . Duel terminates immediately; Team scores 1 point.
- Case 2 (Premature Foul): Neither agent has `has_token == True`, but one agent declares an intentional tag/push. Duel terminates; victim's team scores 1 point.
R5. Successful Base Escape:
- If with `has_token == True` reaches : Duel terminates; Team 0 scores 1 point.
- If with `has_token == True` reaches : Duel terminates; Team 1 scores 1 point.
R6. [gap-fill] Standoff Timeout: If neither agent snatches the token within 100 ticks, the referee drops the handkerchief, declaring a neutral standoff (0 points awarded to both teams), and calls the next number.
R7. Win Condition:
- The first team to accumulate 10 match points wins the overall game.
## Edge cases
- Tie at the end condition: If both agents execute `snatch()` on the exact same tick, [gap-fill] a coin flip awards the token to one agent; the other becomes the immediate tagger.
- A player with no legal move: Waiting / holding position is always legal.
- Repetition / cycles: Bounded by R6 (hard 100-tick limit per duel).
- Simultaneous claims on the same resource or position: Handled by R3/R4 and the simultaneous snatch coin flip.
- Player elimination mid-round; turn order repair: Standard game uses point scoring (no elimination); rounds cycle sequentially through numbers .
- Resource exhaustion: Points accumulate monotonically toward 10.
- Termination guarantee: Enforced by the finite duel timer (100 ticks) and bounded match point cap.
## State representation
- Persistent state:
- `team_scores`: Array of 2 integers `[score_team0, score_team1]`.
- `active_pair`: Tuple of agent IDs `(id_0, id_1)`.
- `round_number`: Integer counter.
- Per-tick ephemeral state:
- `pos_0`: Integer coordinate .
- `pos_1`: Integer coordinate .
- `token_holder`: `None`, 0, or 1.
- `duel_tick`: Integer .
- Observable information: Fully observable by both dueling agents.
## Action space
- `move(dx)`: .
- `snatch()`: Valid when and `token_holder == None`.
## Playability notes
- Single-player-vs-AI: Exceptional 1v1 asymmetric game-theoretic model (feinting, baiting, reaction delay, and sprint threshold optimization).
- Group games (n>4): Reduces naturally to a tournament of 1v1 duels across team lineups.
- Physical-skill games: Physical reaction time and sprint acceleration mapped to discrete tick movement choices and adjacency checks.
- What is lost: Physical agility, feints with shoulders/eyes, and team shouting from the sidelines.
## Implementation hints
- Implement as a simple 1D discrete environment.
- Optimal agent policy requires computing mixed strategy Nash equilibrium at the standoff coordinate.
Download as JSON
Download the full game corpus entry for this game as structured JSON.
Download JSONCite this page
The Handkerchief (Greek Handkerchief Snatch) (Greece). (n.d.). Games of Countries. https://gamesofcountries.com/games/gr-greece/mantilaki
@misc{gamesofcountries-gr-greece-mantilaki,
title = {The Handkerchief (Greek Handkerchief Snatch) (Greece)},
year = {n.d.},
publisher = {Games of Countries},
url = {https://gamesofcountries.com/games/gr-greece/mantilaki}
}