Self-Hosting

Run your own ARES instance on your infrastructure. This guide covers local development setup, production deployment, and configuration options.

Prerequisites

Optional, depending on your provider configuration:

Quick Start

1. Clone the Repository

git clone https://github.com/dirmacs/ares
cd ares

2. Set Up the Database

Create a PostgreSQL database for ARES:

createdb ares

ARES runs migrations automatically on startup. No manual schema setup is required.

3. Create Configuration

Copy the example config and customize it:

cp ares.example.toml ares.toml

Edit ares.toml to configure your providers and models. At minimum, you need one LLM provider:

[server]
port = 3000

[database]
url = "postgres://localhost/ares"

[[providers]]
name = "groq"
type = "openai"
base_url = "https://api.groq.com/openai/v1"
api_key_env = "GROQ_API_KEY"

[[providers.models]]
id = "llama-3.3-70b-versatile"
name = "llama-3.3-70b"
context_length = 131072

4. Set Environment Variables

export DATABASE_URL="postgres://localhost/ares"
export JWT_SECRET="your-secret-key-at-least-32-characters-long"
export API_KEY="your-admin-api-secret"
export GROQ_API_KEY="gsk_..."

5. Build

cargo build --release --features openai,postgres,mcp

See Feature Flags for all available options.

6. Run

./target/release/ares-server

7. Verify

curl http://localhost:3000/health

You should receive a 200 OK response. ARES is running.

Feature Flags

ARES uses Cargo feature flags to control which capabilities are compiled into the binary. This keeps the binary lean — only include what you need.

Build Examples

Minimal build (Groq only):

cargo build --release --no-default-features --features openai,postgres

Full build (all providers):

cargo build --release --features openai,anthropic,ollama,postgres,mcp,ares-vector

Production build (recommended for VPS deployment):

cargo build --release --no-default-features --features openai,postgres,mcp

Production Deployment

systemd Service

Create a systemd unit file at /etc/systemd/system/ares.service:

[Unit]
Description=ARES AI Agent Platform
After=network.target postgresql.service
Wants=postgresql.service

[Service]
Type=simple
User=ares
Group=ares
WorkingDirectory=/opt/ares
ExecStart=/opt/ares/target/release/ares-server
Restart=on-failure
RestartSec=5
Environment=DATABASE_URL=postgres://dirmacs:password@localhost/ares
Environment=JWT_SECRET=your-production-jwt-secret
Environment=API_KEY=your-admin-secret
Environment=GROQ_API_KEY=gsk_...
Environment=RUST_LOG=info

[Install]
WantedBy=multi-user.target

Enable and start:

sudo systemctl enable ares
sudo systemctl start ares
sudo systemctl status ares

View logs:

journalctl -u ares -f

Caddy Reverse Proxy

Caddy provides automatic HTTPS with Let’s Encrypt. Create a Caddyfile:

api.ares.yourdomain.com {
    reverse_proxy localhost:3000
}

Start Caddy:

sudo systemctl enable caddy
sudo systemctl start caddy

Caddy automatically provisions and renews TLS certificates. No manual certificate management is needed.

PostgreSQL Setup

For production, create a dedicated database user:

CREATE USER ares WITH PASSWORD 'strong-password-here';
CREATE DATABASE ares OWNER ares;

Update your DATABASE_URL accordingly:

DATABASE_URL=postgres://ares:strong-password-here@localhost/ares

Configuration Reference

The ares.toml file is the primary configuration file. It controls server settings, providers, models, and agent definitions.

Server Section

[server]
port = 3000          # HTTP port (overrides PORT env var)
host = "0.0.0.0"     # Bind address

Database Section

[database]
url = "postgres://ares:password@localhost/ares"
max_connections = 10

Provider Section

Each provider is defined as a [[providers]] entry:

[[providers]]
name = "groq"
type = "openai"
base_url = "https://api.groq.com/openai/v1"
api_key_env = "GROQ_API_KEY"

[[providers.models]]
id = "llama-3.3-70b-versatile"
name = "llama-3.3-70b"
context_length = 131072

[[providers.models]]
id = "llama-3.1-8b-instant"
name = "llama-3.1-8b"
context_length = 131072

[[providers]]
name = "anthropic"
type = "anthropic"
api_key_env = "ANTHROPIC_API_KEY"

[[providers.models]]
id = "claude-3-5-sonnet-20241022"
name = "claude-3.5-sonnet"
context_length = 200000

[[providers]]
name = "local"
type = "ollama"
base_url = "http://localhost:11434"

[[providers.models]]
id = "mistral"
name = "mistral-7b"
context_length = 32768

Agent Section

Static agents can be defined in the config file:

[[agents]]
name = "general-assistant"
model = "llama-3.3-70b"
system_prompt = "You are a helpful assistant."
tools = ["calculator", "web_search"]
max_tokens = 4096

For tenant-specific agents, use the Admin API instead of config file definitions.

Updating

To update a running ARES instance:

cd /opt/ares
git pull origin main
cargo build --release --no-default-features --features openai,postgres,mcp
sudo systemctl restart ares

Database migrations run automatically on startup. No manual migration steps are needed.

Troubleshooting

Port already in use:

Error: Address already in use (os error 98)

Another process is using port 3000. Either stop it or change the port in ares.toml.

Database connection failed:

Error: error communicating with database

Verify PostgreSQL is running and your DATABASE_URL is correct. Check that the database user has permissions on the database.

Provider API key missing:

Error: Environment variable GROQ_API_KEY not set

Set the required API key environment variable, or remove the provider from ares.toml if you do not need it.

JWT secret too short:

Error: JWT_SECRET must be at least 32 characters

Use a longer secret. Generate one with: openssl rand -hex 32