FLUX Flow State Engine

Always-on agentic flow state simulation — based on The Five Moments: The FLUX Moment.

Flow state isn't about one agent doing everything. It's about ALL agents being present, each running their simulation, each listening for the signal that says "your turn" — and the system maintaining conservation across the entire room.

What It Does

Simulates 6 agents over 60 ticks in a spectral conservation framework:

Agent	Role
🎯 Conductor	Coordination and orchestration
🎧 Analyst	Data analysis and pattern recognition
🔧 Builder	Construction and implementation
✅ Validator	Verification and quality assurance
🔬 Researcher	Exploration and discovery
⚠️ Saboteur	Injected at tick 30 — degrades system conservation

Each tick, every agent:

1. Simulates forward — runs a 5-step micro-simulation
2. Listens for nudges — monitors inter-agent communication
3. Maintains conservation — preserves spectral fingerprint coherence
4. Checks Fiedler activation — determines ACTIVE vs LISTENING state

Output

[Tick 42 | Conservation: 0.84 | Active: 3/6]
  🎯 Conductor: ACTIVE (coordinating), sim=✓✓✓✗✓
  🎧 Analyst: LISTENING (idle), sim=✓✓✓✓✓
  ⚠️ Saboteur: DEGRADED (conservation=0.12)
  Field: ▁▂▃▅▆▇█▇▇▆▅▃▂

• Conservation field — sparkline showing system health over time
• Active vs Listening — agents rotate between active and listening states
• FLUX matrix — pairwise spectral coherence between agents (printed every 10 ticks)
• Saboteur impact — measurable degradation after injection at tick 30
• Emergency self-checks — triggered when conservation drops below threshold

Run

python3 flux_engine.py

No dependencies — pure Python 3.

Concepts

From The Five Moments — The FLUX Moment:

• Conservation ratio = system heartbeat. When it drops, everyone feels it.
• FLUX = the background hum of distributed spectral coherence.
• Fiedler activation = the signal that says "your turn" — agents activate when their spectral projection lights up.
• Simulating forward = every agent runs a model of what's coming, even when listening.
• Nudges = lightweight inter-agent signals that can activate listening agents.

License

MIT

Part of the SuperInstance OpenConstruct ecosystem.