Turning Claude Code into an Enterprise AI Operating System with EvoNexus

Running AI in an enterprise often leads to chaotic processes and a fragmented toolset. While Claude Code is powerful, it operates as a solo agent. EvoNexus addresses this by upgrading Claude Code into a comprehensive AI operating system that can manage an entire company's workflows.

Project Overview

The open‑source project (GitHub: https://github.com/EvolutionAPI/evo-nexus) is built on the Claude Code CLI protocol and transforms a single AI tool into an "AI team".

It ships with 38 dedicated agents divided into two dimensions:

17 business agents covering operations, finance, community, marketing, HR, legal, product, data, customer support, strategy, sales, and other enterprise functions.

21 engineering agents handling architecture, code review, testing, security, design, orchestration, retrospectives, and more; these originate from the oh‑my‑claudecode project and add native Helm and Mirror agents.

Engineering agents follow a six‑stage pipeline: planning → review → implementation → testing → deployment → retrospective.

Core Highlights

Multi‑backend support : By default it runs on Anthropic’s native Claude CLI, but switching to OpenRouter (200+ models), OpenAI, Gemini, AWS Bedrock, Vertex AI, or Codex Auth only requires installing a binary called OpenClaude —no code changes are needed.

Markdown‑First architecture : Each agent is a .md file containing system prompts. No SDK or plugin runtime is required; adding a new agent is as simple as dropping a Markdown file into a directory, greatly lowering customization barriers.

175+ Skills : Skills span finance, community, engineering, data, legal, HR, and operations, all defined as reusable Markdown files.

19 integrations : Direct connections to Google Calendar, Gmail, GitHub, Linear, Discord, Telegram, Stripe, YouTube, Instagram, LinkedIn, Notion, Obsidian, and others via MCP protocol or APIs.

What It Can Do

The official documentation lists typical scenarios:

Every morning an AI agent generates an operations briefing with financial data, community health metrics, and social‑media performance, consolidating everything on a unified dashboard.

For engineering, a post‑commit review agent checks security, code quality, and best practices, then suggests improvements.

In community operations, Discord and Telegram messages are automatically processed for moderation, Q&A, and escalation.

Financially, Stripe revenue data and Brazilian ERP systems such as Omie and Bling can be pulled to produce automated reports.

How to Run

The recommended one‑command installation: npx @evoapi/evo-nexus This command downloads an interactive installer, checks dependencies, asks for company information, selects an AI provider and integration modules, then generates configuration files and builds the dashboard frontend.

Alternatively, you can clone the repository manually:

git clone --depth 1 https://github.com/EvolutionAPI/evo-nexus.git
cd evo-nexus
make setup

Installation requires Claude Code, Python 3.11+, Node.js 18+, and uv. The installer verifies each dependency.

After setup, entering /oracle in Claude Code launches the official entry point, guiding initial configuration and producing a phased activation plan.

Who Should Use It

According to the official description, the system targets:

Technical founders who need an AI assistant that handles both engineering and daily operations.

Small‑to‑mid‑size enterprises that want to automate finance, operations, customer support, and community functions without purchasing a suite of SaaS tools.

Users with custom AI workflow requirements; the Markdown‑First design lets them modify agent behavior without writing code.

Caveats

EvoNexus is an unofficial third‑party project with no direct affiliation to Anthropic; users must provide their own Claude Code installation and credentials.

The system runs locally in a CLI environment, meaning data never leaves the user’s infrastructure—a privacy advantage.

Among the 19 integrations, several are Brazil‑specific ERP and payment tools (Omie, Bling, Asaas); their usefulness is limited outside Brazil.

In summary, EvoNexus expands Claude Code from a single code‑assistant into a full‑featured enterprise AI operating system, offering 38 agents, 175+ skills, 19 integrations, all running locally with a low‑code Markdown‑First customization model, ideal for technically‑savvy founders or small teams building automated workflows.