APEX — Command-Line Operating System for Autonomous Business Units

Welcome to APEX, the central, production-grade system runtime developed by AXIOM LLC. APEX is a minimalist command-line operating system designed to orchestrate and run deterministic Autonomous Business Units (ABUs) on standard POSIX environments.

By consolidating previously fragmented repositories into a single unified monorepo, APEX eliminates version drift, enforces strict schema invariance, and runs parallel workloads with hard OS-level process isolation.

1. The 18 Autonomous Business Unit (ABU) Verticals

APEX includes 18 production-ready business automation templates located inside the templates workspace. These are not trivial examples; they are complete, functional shell workflows designed to run automated workloads, manage risks, and drive recurring revenue:

2. Core System Invariants & Guardrails

Every process and component inside this monorepo strictly complies with these four architectural guardrails:

1. Pure Functional Transition Kernels: The core execution engine (apex/core/loop.py) operates as a pure function: run(Task, Config, Registry) -> State. No global variables, hidden side-effects, or implicit mutable states are permitted.
2. OS-Level Process Isolation: Multi-agent swarm workers (apex/core/swarm.py) run as isolated operating system processes via subprocess.Popen. A thread crash or infinite loop inside a worker cannot propagate and crash the parent master thread.
3. Two-Stage Paranoid Security Gate: Before any generated plan enters execution, it is checked by local, deterministic Regex prefilters (intercepting commands like rm -rf / or chmod 777), followed by a strict security audit conducted in-process by gemini-3.1-flash-lite.
4. Asymmetric Embedding Precision: The integrated RAG engine (apex/core/rag.py) utilizes asymmetric embedding task configurations (RETRIEVAL_DOCUMENT vs RETRIEVAL_QUERY) to maximize semantic vector search precision.

3. Quickstart & Installation

Prerequisites

• Python 3.11 or 3.12 (Unix-based operating system recommended for local timeout mechanics).
• A valid Gemini API key.

Step 1: Install from PyPI

Install the package directly into your virtual environment:

python3 -m venv .venv
source .venv/bin/activate
pip install --upgrade pip
pip install axiom-apex

Step 2: Configure Keys

export GEMINI_API_KEY="your_api_key_here"

Step 3: Run Your First Task

Submit a direct task straight to the compiler:

apex "search for error traces in /var/log/syslog and count their occurrences"

Step 4: Run the Complete Regression Suite

pytest apex/tests/ -q

4. Contact & Registry

• Company: axiom.co@proton.me
• Development: axiom.de@proton.me
• Web Surface: axiom-llc.github.io