Installation

Xagent runs several ways, from a single command on your laptop to a full multi-user deployment. Pick the option that matches your needs.

pip / uv (single machine)

The lightest option — no Docker and no external services. The published package bundles the web UI and defaults to SQLite, so one command runs everything. Requires Python 3.11+.

uv tool install xagent               # isolated, recommended
# or into a virtual environment (PEP 668 on Python 3.11+):
python -m venv .venv && source .venv/bin/activate
pip install xagent

Start the server with the xagent-web command, then open http://127.0.0.1:8000:

xagent-web                              # http://127.0.0.1:8000
xagent-web --host 0.0.0.0 --port 8080   # custom host / port
python -m xagent.web                    # equivalent module form

Or run it without installing, using uvx:

uvx --from xagent xagent-web

Where data lives

Single-machine data is stored under ~/.xagent (SQLite) — no PostgreSQL or Redis required. Override the location with XAGENT_STORAGE_ROOT. For multi-user or team deployments, use Docker Compose below.

Docker Compose (recommended for teams)

The fastest path to a shared, multi-user deployment:

git clone https://github.com/xorbitsai/xagent.git
cd xagent
cp example.env .env
docker compose up -d

Then open http://localhost:80 and create your admin account at /setup.

Manual Installation

For full control over the deployment, run the backend and frontend yourself from a source checkout. Requires Python 3.11+, plus PostgreSQL and Redis (the official Compose stack uses PostgreSQL 16 and Redis 7).

git clone https://github.com/xorbitsai/xagent.git
cd xagent
pip install -e .                 # install from source (pyproject.toml)
cp example.env .env              # edit with your settings
python -m xagent.web             # start the backend on port 8000

The frontend is a separate app:

cd frontend
npm install
npm run dev

System Requirements

ResourceMinimumRecommended
CPU4 cores8+ cores
RAM8 GB16+ GB
Disk20 GB50+ GB SSD

Configure a model first

Before creating tasks, add an LLM provider under Models, or set a provider key in .env (e.g. OPENAI_API_KEY, ANTHROPIC_API_KEY). See LLM Models.

Next Steps