Installing Governance Layer: Setup and Dependencies

The Governance Layer is a Python 3.10+ package distributed as an editable source install. It has no private index dependencies — everything comes from PyPI. The steps below take you from a fresh clone to a running speaker demo, with optional extras for reinforcement-learning experiments.

Clone the repository

Clone the repo and change into the project directory.

git clone https://github.com/xcoder-es/governance-layer.git && cd governance-layer

Install the package

Choose between a standard pip editable install or the faster uv workflow. Both install the same core dependencies.

pip install -e .

The core install brings in all runtime dependencies: gymnasium, neo4j, streamlit, altair, pytest, pytest-cov, pandas, certifi, and hypothesis.

Install optional extras (if needed)

Two optional extras are available for RL and grid-world experiment support.

[rl] — Reinforcement Learning

Pulls in torch>=2.0 and stable-baselines3>=2.3.

pip install -e ".[rl]"

[minigrid] — MiniGrid Environments

Pulls in minigrid>=3.1 and pygame-ce>=2.4.

pip install -e ".[minigrid]"

You can combine extras in a single command:

pip install -e ".[rl,minigrid]"

Configure environment variables

An .env.example file is included at the project root with Neo4j Aura placeholders. If you plan to use the persistent ontology backend, copy it and fill in your credentials.

cp .env.example .env

Then edit .env with your Neo4j connection details:

NEO4J_URI=neo4j+s://your-instance.databases.neo4j.io
NEO4J_USER=neo4j
NEO4J_PASSWORD=your-password-here

The .env file is gitignored. If you skip this step the package still works — the dashboard automatically falls back to the in-memory MemoryBackend.

Verify the installation

Run the Neural Parliament speaker demo to confirm everything is wired up correctly.

python -m src.governance.runner speaker

A successful run prints a step-by-step trace of the parliament deliberating on a sequence of proposals. If you see output without import errors, the install is complete.

Core dependencies

All of the following are automatically installed as part of the base pip install -e . or uv sync step.

Package	Minimum version	Purpose
`gymnasium`	`>=0.29`	RL environment interface
`neo4j`	`>=5.20`	Optional persistent ontology backend
`streamlit`	`>=1.35`	Research dashboard UI
`altair`	`>=5.4`	Chart rendering in dashboard
`pytest`	`>=8.0`	Test runner
`pytest-cov`	`>=5.0`	Coverage reporting
`pandas`	`>=2.0`	Data handling in benchmarks
`certifi`	`>=2024.0`	SSL certificates for Neo4j Aura
`hypothesis`	`>=6.161`	Property-based and fuzz tests

Python version support

The package requires Python 3.10 or later. The CI matrix tests against 3.10, 3.11, and 3.12. Python 3.9 and below are not supported because the codebase uses match/case statements and modern union type syntax.

Introduction

Core Framework

Running Experiments

Configuration & Integration

Installing Governance Layer: Setup and Dependencies

[rl] — Reinforcement Learning

[minigrid] — MiniGrid Environments

Core dependencies

Python version support

Build docs developers (and LLMs) love

Introduction

Core Framework

Running Experiments

Configuration & Integration

Documentation Index

[rl] — Reinforcement Learning

[minigrid] — MiniGrid Environments

​Core dependencies

​Python version support

Build docs developers (and LLMs) love

Core dependencies

Python version support