💻 Run Charles locally

The fastest way to try Charles — run it directly on your own computer. It runs while your machine is on; close your laptop and it pauses. Great for testing before committing to a 24/7 server.

1 · Install Python 3.11 (if you don't have it)

Download from python.org — during install, check "Add Python to PATH". Mac users can also use brew install python@3.11.

2 · Unzip the bundle

Extract charles-customer.zip to a folder you'll remember, e.g. Desktop/charles.

3 · Add your API keys

In the charles folder, copy .env.example to .env and fill in your keys:

ALPACA_API_KEY=your_key
ALPACA_SECRET_KEY=your_secret
GROQ_API_KEY=your_groq_key
TELEGRAM_BOT_TOKEN=your_token
TELEGRAM_CHAT_ID=your_chat_id

4 · Install dependencies & run

# from inside the charles folder
pip install -r requirements.txt
python main.py

A setup wizard opens in your browser. Fill in any remaining settings and click Launch. The dashboard opens at http://localhost:8501. First start can take a few minutes while it loads models and connects.

01 Create your server

Log in to digitalocean.com → Create → Droplets, and choose:

Click Create Droplet, wait ~60 seconds, and copy the IPv4 address it gives you. That IP is your server's identity everywhere below — and the address Charles will register on your access whitelist.

💡 New to DigitalOcean? Sign up with this referral link to get $200 in free credit — that covers ~8 months of server costs, making Charles just $19/month all-in during that period.

02 Connect to the server

On your own machine, open PowerShell and SSH in. Replace YOUR_SERVER_IP with the address from step 1.

ssh root@YOUR_SERVER_IP

Type yes to accept the fingerprint, then your droplet password. Your prompt becomes root@charles:~# — you're now inside the server.

03 Run the deploy script

The deploy script installs everything Charles needs — Miniconda + Python 3.11, the AI runtime, all packages, firewall rules, the auto-start service, and the charles-* commands. First copy it up from your PC, then run it on the server.

# push the deploy script to the server
scp deploy.sh root@YOUR_SERVER_IP:~/deploy.sh

chmod +x deploy.sh
bash deploy.sh

Takes 10–15 minutes. When it's done you'll see the “Deployment Complete” banner with your management commands.

04 Extract the bot onto the server

Now push the bot bundle up and unpack it. You'll do the upload from PowerShell and the extract in bash.

Upload the zip bundle (one clean transfer):

# from the folder that contains charles-customer.zip
scp charles-customer.zip root@YOUR_SERVER_IP:~/charles-customer.zip

Unzip into the bot's home directory and clean up:

mkdir -p ~/charles
apt-get install -y unzip      # if not already present
unzip -o ~/charles-customer.zip -d ~/charles
rm ~/charles-customer.zip
ls ~/charles                  # confirm main.py, brain.py, dashboard.py … are there

Prefer drag-and-drop? FileZilla works too — Host YOUR_SERVER_IP, user root, port 22, upload into /root/charles/. The bash unzip step is only needed if you uploaded a zip.

05 Add your API keys

Charles reads its credentials from a .env file. Create it on the server:

nano ~/charles/.env

Paste your keys, then save with Ctrl+X → Y → Enter:

ALPACA_API_KEY=your_key
ALPACA_SECRET_KEY=your_secret
TELEGRAM_BOT_TOKEN=your_token
TELEGRAM_CHAT_ID=your_chat_id
GROQ_API_KEY=your_groq_key

💡 Live trading: Charles runs in paper mode by default. To enable live trading, create an Alpaca live account at alpaca.markets, get your live API keys, and replace the paper keys in your .env file. No other changes needed.

06 Run Charles

charles-start

✅ Charles started
📊 Dashboard: http://YOUR_SERVER_IP:8501
📋 Logs: charles-logs

Open http://YOUR_SERVER_IP:8501 in any browser to watch the live dashboard. The bot now runs 24/7 and restarts itself on reboot. That's it — Charles is live.

⌘ Command reference

Everything you'll ever need to operate Charles, from your bash terminal on the server:

Updating the bot later: re-upload your files from PowerShell with scp (step 4), then run charles-restart on the server.

! Troubleshooting