Skip to content

Getting started

Prerequisites

  • Python 3.10 or later (3.13 recommended).
  • PostgreSQL 16 with the pgvector extension (optional for core tests, required for the daemon).
  • make.

Install

git clone https://github.com/yohanpoul/etz-chaim-ai.git
cd etz-chaim-ai
make install

This creates a virtual environment at .venv/, installs the project in editable mode with development and documentation extras, and wires pre-commit hooks.

Run the core tests

make test-core

You should see around 200 tests green across the core modules.

Database setup (optional for v0.1.0)

If you want to run the daemon or the runtime validation cycle :

createdb etz_chaim
psql etz_chaim -c "CREATE EXTENSION IF NOT EXISTS pgvector;"
psql etz_chaim < scripts/init_schema.sql

Then configure the connection via the ETZ_CHAIM_DB_URL environment variable, for example :

export ETZ_CHAIM_DB_URL="postgresql://postgres@localhost:5432/etz_chaim"

Run the demo cycle

make demo

This runs scripts/force_probe_cycle.py, which :

  1. Takes a snapshot of the configuration state.
  2. Runs one probe orchestrator cycle over the current state.
  3. Verifies that no row of the configuration state was written (layered composition compliance).
  4. Reports any rectifiers emitted.

Expected output ends with Verdict : ✓ FIX HOLDS.

Explore the public API

from etzchaim import initiate

# Plug your LLM into Etz Chaim AI
agent = initiate(llm="claude-opus-4")

# Run a query through the cognitive operating system
response = agent.query("What are your typical failure modes?")

For corpus exploration and advanced usage, see etzchaim --explain-origin or docs/advanced.md.

Next steps