Meet OpenAcme: an AI workforce you run yourself

Most AI tools give you a chat box. You type, it answers, you type again. Close the tab and the thread is gone. OpenAcme is built around a different idea: a workforce of agents that work for you, keep their context between sessions, and hand results back when they're done.

You're not a user here. You're in charge.

A roster, not a thread

In OpenAcme, every agent is a coworker you can describe in a sentence. It has a name, a role (what it does, written the way you'd brief a teammate), and a persona (how it behaves). It carries its own model, tools, skills, memory, and workspace — independent from everyone else on the roster.

That separation is the whole point. An analyst and an engineer shouldn't share the same browser logins, the same memory, or the same model. Each agent is its own thing. You add one, retire one, or swap a model without touching the rest.

Hand off work and walk away

Agents don't just answer questions — they take tasks. You write a task the way you'd brief a coworker: a title and what "done" looks like. It lands on a shared board, and the dispatcher wakes the right agent to pick it up, usually within the minute.

• Tasks can recur — a finished daily task re-opens itself for tomorrow.
• Tasks can depend on each other — dependents wait their turn.
• Results come back as comments, with the full event log of what happened.

So you can assign three things on your way out and read the answers when you're back, instead of babysitting a prompt.

Everything a coworker needs

A chat window with plugins isn't a coworker. An agent that gets real work done needs its own working setup:

• Memory — corrections, preferences, and context saved to a Markdown file per agent, carried into every future session.
• Skills — teachable capabilities in Markdown. Install them from GitHub and marketplaces, or write your own.
• A browser — each agent gets its own browser session, with its own logins and cookies. One agent's session never leaks into another's.
• Any model — mix providers across the roster. Local models work too.
• MCP — wire up any Model Context Protocol server; its tools show up namespaced per agent or shared across the workforce.
• Teams — send work to a team and the manager triages it: claim, reassign, or split.

It runs where you work

OpenAcme is self-hosted. There is no OpenAcme cloud sitting between you and your data — your prompts and files go to the model provider you chose and nowhere else.

• On your laptop: install, sign in, done. Everything lives in ~/.openacme — plain Markdown and SQLite you can open, read, and back up.
• On a shared server: run openacme expose and your whole team works with the same workforce, behind one URL, with real accounts.

The web console and the CLI are the same product. Anything you can do in the browser, you can do from the terminal — including openacme chat, which runs an agent in-process with no server at all.

Get started

npm install -g @openacme/cli && openacme setup

That walks you through connecting a provider — a Claude or ChatGPT subscription, or an API key — and you're ready to hire your first agent. The Quickstart takes it from there.

This is the first of these notes. We'll use the blog for design notes and the things we learn building this, and the changelog for what ships. Welcome aboard.