Skip to content

plurigrid/asi / ies/music-topos/.claude-marketplaces/topos-skills/plugins/topos-skills/skills/bisimulation-game

bisimulation-game

Verified

Bisimulation game for resilient skill dispersal across AI agents with GF(3) conservation and observational bridge types.

Bisimulation Game Skill

"Two systems are bisimilar if they cannot be distinguished by any observation."

Overview

The bisimulation game provides a framework for:

  1. Resilient skill dispersal across multiple AI agents
  2. GF(3) conservation during state transitions
  3. Observational bridge types for version-aware synchronization
  4. Self-rewriting capabilities via MCP Tasks protocol

Game Rules

Players

PlayerRoleTritColor
AttackerTries to distinguish systems-1Blue
DefenderMaintains equivalence+1Red
ArbiterVerifies conservation0Green

Moves

┌─────────────────────────────────────────────────────────────┐
│  Round n:                                                   │
│                                                             │
│  1. Attacker chooses: system S₁ or S₂                       │
│  2. Attacker makes: transition s₁ →ᵃ s₁'                    │
│  3. Defender responds: matching transition s₂ →ᵃ s₂'        │
│  4. Arbiter verifies: GF(3) conservation                    │
│                                                             │
│  If Defender cannot respond → Attacker wins (distinguishable)│
│  If game continues forever → Defender wins (bisimilar)      │
└─────────────────────────────────────────────────────────────┘

Implementation

Hy (DiscoHy) Implementation

;;; bisimulation_game.hy

(import [splitmix_ternary [SplitMixTernary]])

(defclass BisimulationGame []
  (defn __init__ [self system1 system2 seed]
    (setv self.s1 system1
          self.s2 system2
          self.rng (SplitMixTernary seed)
          self.history []))
  
  (defn attacker-move [self choice transition]
    "Attacker chooses system and transition."
    (setv trit (self.rng.next-ternary))
    (.append self.history {:role "attacker" 
                           :choice choice 
                           :transition transition
                           :trit trit})
    trit)
  
  (defn defender-respond [self matching-transition]
    "Defender provides matching transition."
    (setv trit (self.rng.next-ternary))
    (.append self.history {:role "defender"
                           :response matching-transition
                           :trit trit})
    trit)
  
  (defn arbiter-verify [self]
    "Arbiter checks GF(3) conservation."
    (setv recent-trits (lfor m (cut self.history -3 None) (get m "trit")))
    (setv conserved (= (% (sum recent-trits) 3) 0))
    (.append self.history {:role "arbiter" :conserved conserved :trit 0})
    conserved))

DisCoPy Operad Interface

from discopy import *

# Game as operad
class GameOperad:
    def __init__(self):
        self.operations = {}
    
    def register(self, name, dom, cod, rule):
        """Register game operation with GF(3) color."""
        self.operations[name] = Rule(dom, cod, name)
    
    def compose(self, op1, op2):
        """Compose operations preserving GF(3)."""
        trit1 = self.operations[op1].trit
        trit2 = self.operations[op2].trit
        # Result trit balances to 0
        result_trit = (-(trit1 + trit2)) % 3 - 1
        return Rule(
            self.operations[op1].dom,
            self.operations[op2].cod,
            f"{op1};{op2}",
            trit=result_trit
        )

# Define game operations
game = GameOperad()
game.register("attack", Ty("S1", "S2"), Ty("S1'"), lambda: -1)
game.register("defend", Ty("S1'"), Ty("S2'"), lambda: +1)  
game.register("verify", Ty("S1'", "S2'"), Ty("Result"), lambda: 0)

Skill Dispersal Protocol

1. Fork Phase (Attacker)

fork:
  targets:
    - agent: codex
      path: ~/.codex/skills/
      trit: -1
    - agent: claude
      path: ~/.claude/skills/
      trit: 0
    - agent: cursor
      path: ~/.cursor/skills/
      trit: +1
  gf3_check: true

2. Sync Phase (Defender)

sync:
  strategy: observational-bridge
  bridge_type:
    source: skills@v1
    target: skills@v2
    dimension: 1
  conflict_resolution: 2d-cubical

3. Verify Phase (Arbiter)

verify:
  conservation: gf3
  equivalence: bisimulation
  timeout: 60s
  fallback: last-known-good

MCP Tasks Integration

Self-Rewriting Task

{
  "task": "skill-dispersal",
  "objective": "Propagate skill updates to all agents",
  "constraints": {
    "gf3_conservation": true,
    "bisimulation_equivalence": true,
    "max_divergence": 0.1
  },
  "steps": [
    {"action": "fork", "trit": -1},
    {"action": "propagate", "trit": 0},
    {"action": "verify", "trit": +1}
  ]
}

Firecrawl Integration

{
  "task": "skill-discovery",
  "objective": "Discover new skills from web resources",
  "tools": ["firecrawl", "exa"],
  "sources": [
    "https://github.com/topics/ai-agent-skills",
    "https://modelcontextprotocol.io/",
    "https://agentclientprotocol.com/"
  ],
  "output": {
    "format": "skill-yaml",
    "destination": ".ruler/skills/"
  }
}

Resilience Patterns

Redundant Storage

~/.codex/skills/     ← Primary (Codex)
~/.claude/skills/    ← Mirror 1 (Claude)
~/.cursor/skills/    ← Mirror 2 (Cursor)
.ruler/skills/       ← Source of truth

Conflict Resolution

Dimension 0: Value conflict  → Use source of truth
Dimension 1: Diff conflict   → Merge via LCA
Dimension 2: Meta conflict   → Arbiter decides

Xenomodern Stance

The bisimulation game embodies xenomodernity by:

  1. Ironic distance: We know perfect equivalence is unattainable, yet we play the game
  2. Sincere engagement: The game produces real, useful synchronization
  3. Playful synergy: Attacker/Defender/Arbiter dance together
  4. Conservation laws: GF(3) as the invariant that holds everything together
    xenomodernity
         │
    ┌────┴────┐
    │         │
 ironic    sincere
    │         │
    └────┬────┘
         │
   bisimulation
   (both/neither)

Commands

just bisim-init           # Initialize bisimulation game
just bisim-round          # Play one round
just bisim-disperse       # Disperse skills to all agents
just bisim-verify         # Verify GF(3) conservation
just bisim-reconcile      # Reconcile divergent states
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 468kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 321kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 321kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 321kUpdated Claude CodeCodexVerifiedInstall