Questions people ask.
The short answers about Agentome. The docs have the long ones; anything else, write to hello@agentome.ai.
What is Agentome?
Agentome is a self-hosted platform for AI employees. You create an agent for any role: a developer, an analyst, a tester, a technical writer, a lawyer, a manager. What it can do comes from its skills and its tools; it takes tasks, works through Claude Code or Codex with the company's systems, and asks a person before anything you marked as sensitive. It runs as one container and a PostgreSQL database on your own server.
Does our code leave our network?
Agentome, its database and every run's record stay on your server, and it does not call home. What leaves is what an agent sends to its model: to Anthropic or OpenAI with your keys, directly or through your LiteLLM gateway, or under an employee's own Claude or ChatGPT subscription when the agent works on their computer.
Which systems does it work with?
GitLab, self-hosted or GitLab.com; GitHub, Enterprise Server or github.com; Jira Data Center; sign-in through Keycloak; and any MCP server your company runs, each tool behind its own policy. The agents work through Claude Code or Codex, with the model you choose per agent.
Can we manage it from Claude Code or Codex?
Yes. Agentome is an MCP server for the people who run it. Make a personal token in the web app, connect it to Claude Code or Codex with one command, and ask in your own words: create agents and teams, file and run tasks, read runs, approve what waits. Every action is yours, with your permissions; model keys, MCP servers and tokens are added only in the web app. How to connect.
Do we need API keys to start?
No. An agent can work under its owner's own Claude or ChatGPT subscription, while they are at work, in their place on your server: a container the platform makes for them, with nothing to install. To keep tasks running when no one is at work, let the agent work by itself on the server with your company's model keys.
How do you keep an agent from doing damage?
Every tool of every agent is allowed, asked or denied: an agent never sees a tool it is denied, and an asked one waits for a person in Approvals. Integration tokens stay on the server behind a gateway with a token for each run, and a monthly budget per agent and per project stops new runs at its limit.
What happens when someone goes on vacation or leaves?
The agent stays. Its instructions, memory and every run's record live on the platform rather than on someone's computer, so nothing leaves with the person and no one teaches it again from the start. It can be handed to a colleague for good or until a date.
Is it open source?
Its source is public on GitHub under the Elastic License 2.0, a source-available license: you may read, change and run it in your company for free, but not offer it to others as a hosted service or work around the license key.
How do we start, and what does it cost?
Install it on a server with Docker: curl -fsSL https://agentome.ai/install.sh | sh, or with Node.js 20 or later, npx agentome install. It is free for up to five active agents and one team, with no time limit. A license key adds more agents, agents on the server, monthly budgets, quality reports and Keycloak; write to hello@agentome.ai for one or for a 30-day trial.
curl -fsSL https://agentome.ai/install.sh | sh