Minimal Private Deployment

Single-host Docker Compose deployment · Bring up a minimal usable environment in 30 minutes

Goal: bring up a minimal Evose on a single Linux machine using Docker Compose, to validate the product. About 30 minutes.

This is not a production deployment

This guide is for fast validation only. For production, follow the full private deployment guide.

Prerequisites

Resource	Minimum	Recommended
CPU	4 cores	8 cores
Memory	16 GB	32 GB
Disk	200 GB SSD	500 GB SSD
OS	Ubuntu 20.04+ / CentOS 7+	Ubuntu 22.04
Docker	20.10+	24+
Docker Compose	v2	v2

You will also need at least one LLM access credential (OpenAI / Anthropic / Tongyi / ERNIE / self-hosted model API).

Steps

1 · Prepare the environment (5 minutes)

# Verify Docker
docker --version
docker compose version
 
# Create working directory
mkdir -p /opt/evose && cd /opt/evose

2 · Get the deployment package (2 minutes)

# Assumes you have the deployment package (contact sales)
tar -xzf evose-edition-*.tar.gz
cd evose-edition

Deployment package layout:

evose-edition/
├── docker-compose.yml       # Compose file
├── .env.example             # Env var template
├── config/                  # Config directory
└── scripts/                 # Ops scripts

3 · Configure environment variables (5 minutes)

cp .env.example .env
vim .env

Minimum required:

# Domain
EVOSE_DOMAIN=evose.example.com
 
# DB password (generate randomly)
DB_PASSWORD=<generate via openssl rand -hex 16>
REDIS_PASSWORD=<same>
 
# Encryption key (used for credential encryption)
SECRET_KEY=<generate via openssl rand -hex 32>
 
# Initial admin
ADMIN_EMAIL=admin@example.com
ADMIN_PASSWORD=<at least 12-char strong password>
 
# At least one LLM
LLM_PROVIDER=openai
OPENAI_API_KEY=<your key>
OPENAI_BASE_URL=https://api.openai.com/v1

→ Full environment variable list

4 · Start the services (10 minutes)

# Pull images and start
docker compose pull
docker compose up -d
 
# Wait for initialization
docker compose logs -f evose-init
 
# Continue after seeing "Initialization complete"

Health check:

# All services should be healthy
docker compose ps

5 · Access and initialize (5 minutes)

Open a browser and go to http://<server IP>:8080 (or your configured domain)
Log in with ADMIN_EMAIL / ADMIN_PASSWORD from .env
The auto-generated default Organization + default Workspace are ready
Go to Org management · Model platform and confirm your LLM is recognized
Enter the Workspace and follow step 3 of the SaaS 5-minute tutorial (starting at "Create your first Agent from a template")

Verification Checklist

docker compose ps shows all containers healthy
You can log into the admin console in the browser
Model platform shows the configured LLM
You can create and chat with the first Agent in the Workspace

FAQ

Database initialization fails?

Check that DB_PASSWORD does not contain special characters (avoid $, spaces); ensure the disk has enough free space.

Model call returns 401?

Go to Org management · Security · Credentials, confirm the API Key is correct; click "Test connection" on the Model platform.

How to upgrade / back up?

Upgrade procedure
Backup: docker compose exec postgres pg_dump ... (see production docs)

This Is Just a Minimal Deployment

Once it runs, before going to production you'll need:

Required	Document
High availability (dual instances + standalone DB)	HA deployment
Kubernetes orchestration	K8s deployment
SSO integration	SSO config
HTTPS / domain / CDN	System requirements
Backup and upgrade procedures	Upgrade
6-layer defense	Defense in depth

Next Steps

Want to experience the full feature set → First Agent
Want a production deployment → System requirements

Minimal Private Deployment

On this page