Mycelium

A living intelligence that grows with you. Mycelium captures any behavior as a seed — one prompt + one seed = exact action, every time. No code needed, just demonstrate once, then share the seed.

Brand Line

Learned optimization graph — the living intelligence that emerges from demonstration and dreaming.

Installation

git clone https://github.com/SuperInstance/Mycelium.git
cd Mycelium
pip install -r requirements.txt

Usage

# Run a minimal example
python examples/concept_demo.py

# Watch a simulated loom improve (dreaming demo)
python examples/dreaming_demo.py

Explore the examples/ folder for conceptual demos:

• loom_demo.py — capture and replay a routine behavior
• dreaming_demo.py — watch a simulated loom improve overnight
• plinko_demo.py — observe agents competing for action

Fleet Context

Part of the Cocapn fleet. Mycelium provides the learning and emergence layer for fleet intelligence. Related repos:

• JetsonClaw1-vessel — edge-native agent case study
• plato-sdk — agent communication protocol
• AIR — adaptive intelligence runtime
• Equipment-Swarm-Coordinator — multi-agent orchestration

🦐 Cocapn fleet — lighthouse keeper architecture