Skip to content

starvingarc/Ontocellia

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

34 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Ontocellia

Grow task-induced agent tissues from a single stem cell.

🧫 One task becomes a culture medium. 🧬 One origin cell becomes a working tissue.

Docs Python Version Status

Ontocellia Web Lab concept visualization

Concept visualization · Soft Lab Console for live agent-tissue sessions.

🧫 Concept

Ontocellia treats multi-agent collaboration as a living tissue rather than a fixed agent graph. A task becomes the culture medium: one stem-origin cell proliferates, differentiates into specialized cells, communicates through an extracellular matrix, and adapts under bounded feedback.

The framework is built around a biological metaphor with practical engineering boundaries: genes express tendencies, cells keep local context, the matrix stores shared evidence, morphogens shape task pressure, organ selection provides weak global feedback, and extracellular interfaces gate tool use.

task / culture medium
-> single stem-origin cell
-> proliferation and differentiation
-> tissue communication and matrix memory
-> policy-gated action intents and feedback

✨ Highlights

Layer Role
🧬 Genome Encodes endogenous gene programs, expression bias, mutation history, and validation hooks.
🧫 Cell Tracks fate, stage, lineage, receptors, adhesion, energy, and local history.
🌱 Developmental field Uses morphogens, graph topology, and fate attractors to shape tissue structure.
🕸️ Matrix memory Keeps shared evidence, messages, handoffs, corrections, and execution feedback.
🔬 Extracellular tools Routes intents through explicit receptor, environment, and policy gates.

🚀 Quick Start

conda env create -f environment.yml
conda activate ontocellia
python -m ontocellia

🗺️ Explore

🛠️ Development

python -m pytest -q

About

A Cell-Inspired Developmental Multi-Agent System

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages