Skip to main content

​
Getting Started with EvoNexus

​
Choose your install method

Pick the Docker flow if you’re unsure. Keep reading for the CLI flow.

​
Prerequisites (CLI flow only)

  • Claude Code CLI — Install Claude Code
  • Python 3.11+ with uv
  • Node.js 18+ (for the dashboard)
  • API keys for integrations you want to use

​
Installation

​
Option A — Docker (fastest)

Pulls the official multi-arch images and runs the wizard at http://localhost:8080. Full guide: Installing with Docker.
Skip to Start with /oracle after the wizard completes.

​
Option B — npx (CLI flow)

This downloads and runs the interactive setup wizard automatically.

​
Option C — Manual clone

The wizard asks for:
  • Your name and company
  • Timezone and language
  • Which AI provider to use (Anthropic by default; alternatives via OpenClaude)
  • Which agents to enable
  • Which integrations to configure
It generates:
  • config/workspace.yaml — central config
  • config/routines.yaml — routine schedules
  • config/providers.json — active AI provider + backend CLI config
  • .env — API keys (fill in after setup)
  • CLAUDE.md — context file for Claude

​
2. Choose Your AI Provider

The wizard asks which backend should power EvoNexus. Anthropic is the default — if you already have Claude Code authenticated, you don’t need to do anything else. For any other provider (OpenRouter, OpenAI, Gemini, AWS Bedrock, Vertex AI, Codex Auth), EvoNexus uses OpenClaude, a drop-in binary compatible with the Claude CLI protocol. Install it once:
Then select the provider in the wizard (or later from the Providers page in the dashboard) and fill in the keys. The active provider is stored in config/providers.json and both the terminal-server and the ADW runner re-read it on every session spawn — no restart required when switching. See docs/dashboard/providers.md for the full provider reference and docs/reference/env-variables.md for all provider-related env vars.

​
3. Configure API Keys

Edit .env with your keys:
At minimum, you need:
  • No keys required for basic operation (agents, skills work without integrations)
  • DISCORD_BOT_TOKEN — for community monitoring
  • STRIPE_SECRET_KEY — for financial routines
  • Social OAuth keys — via the dashboard Integrations page

​
4. Start the Dashboard

On a VPS (remote): The setup wizard automatically creates a dedicated evonexus system user (Claude Code refuses --dangerously-skip-permissions as root) and installs a systemd service that starts on boot:
You can also install the systemd service manually on an existing installation:
Local (macOS/Linux):
Open http://localhost:8080 — the first run shows a setup wizard where you create your admin account and configure the workspace. Dashboard

​
5. Start Automated Routines

On a VPS, the scheduler runs automatically inside the dashboard service. Locally:
This starts the scheduler that runs routines at their configured times (see config/routines.yaml).

​
6. Use Claude Code — start with /oracle

Open Claude Code in this directory. It reads CLAUDE.md automatically and has access to all agents and skills. First thing to run: /oracle. Oracle is the official entry point — it interviews you about your business, maps workspace capabilities to your pain points, and delivers a phased activation plan through the prod-activation-plan skill (index file + folder-per-phase + file-per-item, each with suggested agent team and pending decisions). You never have to guess the next step.
After the plan is ready, you can invoke individual agents:

​
Next Steps

  • Read Architecture to understand how agents, skills, and routines work together
  • Browse .claude/skills/CLAUDE.md for the full skill index (175+ skills)
  • Check ROUTINES.md for routine documentation
  • Customize config/routines.yaml to adjust schedules