Install Agentivity with Docker

This page gets Agentivity running on your own machine or server. It takes a few minutes.

• A computer or a server that can run Docker: Docker Desktop on Windows or macOS, Docker Engine on Linux. If you do not have it yet, get Docker.
• An API key from an AI model provider. You can add it after the install: see Connect an AI model.

Run this command in a terminal:

docker run -d --name agentivity -p 5005:5005 -v agentivity-data:/app/_data agentivitylabs/agentivity

What the command does:

• -p 5005:5005 makes Agentivity reachable on port 5005 of your machine. If that port is already used, change the first number, for example -p 8080:5005, and open port 8080 instead.
• -v agentivity-data:/app/_data keeps your data in a Docker volume named agentivity-data, so it survives a restart or an update.
• --name agentivity gives the container a name, so you can stop and start it easily.

In your browser, on the machine where Agentivity runs, open:

http://localhost:5005

The Studio opens with a short welcome guide. This is where you build your agents, teams and workflows.

How to know it worked: you see the Studio in your browser. If the page does not load, check that the container is running with docker ps.

Agentivity has no sign-in, no password and no token yet. Anyone who can reach the server can use it. Sign-in will come with the cloud versions.

To try it and to develop, there is nothing to set. By default, the server accepts calls from any web page: a VibeApp or a sample served from another port, another machine or another domain works at once. Calls that do not come from a browser (a script, a mobile application, another server) have always been accepted.

This is easy, and it is not secure. Do it on your own machine, or on a network you trust.

For a real production, do two things.

1. Limit the pages that may call the server to those of your VibeApp. Give their addresses when you start the container, one setting for each address:

   docker run -d --name agentivity -p 5005:5005 -v agentivity-data:/app/_data \
     -e Cors__AllowedOrigins__0=https://app.mydomain.com \
     agentivitylabs/agentivity

   For another address, add Cors__AllowedOrigins__1, then __2, and so on. Without this setting, any website opened by a visitor who can reach your server can call it in the background.

2. Do not expose the server without protection, since it has no sign-in. Put it behind a proxy in HTTPS that controls who gets in, or limit it with a firewall to the people and machines that must use it.

The Studio is not concerned: the server delivers it itself, so it always calls the server from its own address.

• There is no login yet. Anyone who can reach the server can use it. See Who can call the API.

• Your data stays with you. Agents, teams, workflows, tables and settings are stored in the agentivity-data volume. Back it up like any other data. What you send in an AI step goes to the model provider you chose.

• Stop and start. Use docker stop agentivity to stop it and docker start agentivity to start it again. Your data is kept.

• Updating. Pull the new image, remove the container, then run the start command again. The volume keeps your data.

  docker pull agentivitylabs/agentivity
  docker rm -f agentivity

• Licence. The Community edition is source available under the Elastic License 2.0: you can use it for your own business, on your own server, free of charge.

See Fix a failed installation. To keep your data safe and to update, see Update Agentivity and back up your data. Before other people reach your server, read Keep your Agentivity server private.

Connect an AI model, so that your agents have something to think with.