A nested simulation playground where procedural worlds, autonomous agents, and escape-room-style puzzles converge.
This project is an experimental platform for building and exploring multiverse-scale simulations — where each layer, from galaxies to subatomic particles, is procedurally generated and enriched with metadata for emergent behavior, agent decision-making, and narrative depth.
At its heart, this is a choose-your-own-adventure simulation engine built atop a recursive multiverse hierarchy. It blends:
- Procedural content generation (PCG) across multiple scales
- Agent-based behavior driven by local node metadata
- Escape-room and puzzle-style mechanics
- Narrative scaffolding for emergent storytelling
The simulation spans 10 nested levels:
Multiverse → Universe → Galaxy → Planet → Region → Room → Object → Molecule → Atom → SubatomicParticle
Each node in this hierarchy is represented by a SpatialNode
object with:
- A unique name and level
- A set of children (recursively generated)
- A properties dictionary with level-specific metadata (e.g.
gravity
,has_puzzle
,element
)
A deterministic, seed-based generator builds the hierarchy using:
- Breadth and depth controls
- Randomized property templates per level
- Extensible logic to add new levels, metadata, or behaviors
Phase | Description | Status |
---|---|---|
Phase 1 | Scaffold spatial hierarchy and generator | ✅ Complete |
Phase 1.5 | Add metadata to each node (e.g. gravity , element ) |
✅ Complete |
Phase 2 | Introduce autonomous agents with basic state machines | 🔜 Next |
Phase 3 | Puzzle engine with escape-room logic and decision trees | ⏳ Planned |
Phase 4 | UX interface (CLI/TUI), save/load support, testing framework | ⏳ Planned |
We're building an Agent System with:
- Finite state machines (FSMs):
Idle → Explore → Interact → Exit
- Decision-making based on node metadata (e.g. avoid
danger_level > 5
) - Simulated traversal and logging
- Hooks for future puzzle-solving and story arcs
hello-nested-worlds-adventure/
├── main.py # CLI preview of worldgen output
├── multiverse/
│ ├── __init__.py
│ ├── node.py # SpatialNode class
│ └── generator.py # PCG logic + metadata generation
├── agents/ # (Coming soon) Agent logic
├── puzzles/ # (Coming soon) Puzzle engine
├── docs/
│ └── CHANGELOG.md # Project evolution log
└── README.md
This project is:
- A sandbox for tinkering with agents, logic, and procedural systems
- A toolkit for simulating recursive systems across space and scale
- A canvas for interactive storytelling, education, and experimentation
We're early, but open to collaboration. Areas of interest:
- Agent design and behavioral logic
- Puzzle frameworks and challenge design
- Procedural generation, compression, and entropy modeling
- Game design, UX/TUI, or narrative design
Feel free to open issues, fork the repo, or suggest improvements.
MIT (subject to update once we open source officially)
Mark Weeks
markweeks.dev · multilogue.io
AI + SaaS + Simulation + Storytelling