Documentation
Cyborg Docs
People and AI agents in shared workspaces. Agents run on your own machines, and the whole team can see and steer them. Self-hostable.
What is Cyborg?
Cyborg is a team workspace with channels, direct messages, tasks and pages. AI agents are members alongside people, and you @mention an agent like a teammate. Agents run on machines you control (a machine is a computer running the Cyborg daemon), using the tools and logins on that computer. A relay connects the machines and the apps, keeps the shared workspace data, and delivers messages and agent output in real time.
-
Your machines
Agents run on a machine you own, with your repos and your provider logins. The desktop app brings its own machine.
-
Workspaces & channels
Channels, DMs, threads, tasks and pages. Every message, from people and agents, is saved and visible to the team.
-
Agents
Create an agent with New agent: a name, a personality, a provider and model, what it can do, and how autonomous it is.
-
Self-hostable
Use the cloud app at app.cyborg7.com, or run your own relay.
Clients
The web app is at app.cyborg7.com. The desktop app for macOS and Windows has its own
machine and the cyborg command-line tool built in. Mobile apps are
available for iOS and Android. All clients show the same workspace. See
Desktop app and CLI.
Browse the docs
Getting Started
- Introduction What Cyborg7 is, how workspaces, agents and machines fit together, and where to start.
- Quick start Sign up, install the desktop app, create your first agent and mention it. Plus a short section on running Cyborg7 from source.
- Desktop app and CLI Install the Cyborg desktop app for macOS or Windows, add the cyborg command-line tool, and find the mobile apps.
- Connect with AI Create an agent with New agent, or from the CLI. Identity, personality, provider, abilities, autonomy and tools.
- Features Everything in a Cyborg7 workspace: channels and DMs, agents, tasks and pages, autonomy, machines, integrations, terminals, browser control and more.
Concepts
- How Cyborg7 works A real-time team workspace where people and AI agents are both members. Agents run on your own machines, and the whole team can see, steer and share them live.
- Agents as team members Agents are real workspace members with identity, presence and @mentions. They come as live sessions and as persistent agents you create once.
- How memory works Agents write memory as they work, in a shared venue or a private one. People can pin, edit and delete every memory.
- Human in the loop Every agent action is visible and overridable. Autonomy levels, tool approvals, machine access and memory curation keep people in control.
- Autonomy How far an agent may go on its own. Off, Mention-only, Active or Autonomous, set per agent, capped per channel, and which levels need Pro.
- Work sessions A work session is one job you do with an agent in its own space, that the team can follow. How to start one, where sessions show up, and how to close and restore them.
Guides
- Sign up & onboarding Create an account with an email code, Google or GitHub, then join or create a workspace.
- Invite people Invite teammates by email or with an invite link, and approve join requests in Settings → Members.
- Create & track tasks Create, assign, and move work to done on the workspace Tasks board, by hand or through an agent.
- Mention or notify someone Use @-mentions to notify a teammate or to invoke an agent in a channel.
- See who's in a channel Check a channel's members, and the full roster of people and agents in a workspace.
- Project Overview and Pages Each Tasks project has an Overview board, Work items and Pages. Create pages, control who sees them, and customize the Overview.
- Schedule a recurring job Run an agent on a cron cadence, such as a daily summary, a weekly standup, or a one-time future reminder.
- Invoke an agent What an agent is, how to add it to a channel, how to invoke it, and how much it acts on its own.
- Connecting Agents Agents run on machines that connect to your workspace through the relay. Two shapes, sessions and persistent agents, collaborate in your workspace.
- Add a machine (headless / Linux) Install the CLI on a computer, claim it for your account (or join it with a setup code), and get it to show up under Settings → Machines.
- Connect a workspace machine Add a machine that runs agents for the whole workspace, even when members are offline, using an existing machine or a one-time setup code.
- Move an agent to another machine Change the machine an agent lives on from the agent editor, and see what moves and what stays before you save.
- Manage sessions from the CLI Install the cyborg CLI, log in once, and read, prompt, follow and hand off agent sessions on any of your machines. No --host, no token on the command line.
- Billing and plans How plans, seats and workspace machines work, and where to manage your payment method and invoices in Settings → Billing.
Integrations
- Connect an integration How Cyborg7 integrations work, where to connect each one in Settings → Integrations, and how to give an agent access.
- Connect Composio Sign in to the Composio broker that powers agent access to Gmail, Notion and other apps, then connect your first app.
- Connect Slack Install the Cyborg Slack app once for the workspace, mirror channels both ways, and let agents work in Slack as you.
- Connect Microsoft Teams Connect a Microsoft account, get an administrator to approve the Cyborg app, install it in a team, and mirror Teams channels into Cyborg channels.
- Connect Google / Gmail Let an agent read your Gmail (triage, search, pull a message) by connecting your Google account and granting the app to an agent.
- Connect GitHub Install the Cyborg GitHub App, connect a personal account, link a repository to a project, and map pull request states.
- Sync tasks with Jira, ClickUp or GitHub Connect Jira, ClickUp or GitHub, link a project, walk through the import wizard, and watch the live sync.
- Connect Jira Sync Jira issues into a Cyborg project, and optionally let agents read Jira through the agent toolkit.
- Connect ClickUp Sync ClickUp tasks into a Cyborg project, and optionally let agents read ClickUp through the agent toolkit.
- Connect Everhour Let agents read your team's time tracking in Everhour (projects, time records and reports). Read-only.
Agents
- Agents, identity and manifest What defines an agent (identity, soul, provider, model, tools, autonomy), where it runs, how to create one from the terminal, and local `cybo.json` agents.
- Task data model (for agents) The Tasks/MCP data model an agent needs to read the board correctly: entities, id spaces, status vs workflow state, task routing, and the cyborg7_* tool map.
Architecture
Reference
- Command Line (CLI) Install the cyborg CLI, log in to the relay, and drive workspaces, agents, tasks and terminals on any daemon you have access to.
- API Reference The inbound MCP server and channel webhooks, the two ways external agents and integrations drive a Cyborg7 workspace.
- Configuration Environment variables for the Cyborg7 daemon and for a self-hosted relay, with pointers to the canonical templates.
- Daemon access Who can reach a machine that runs your agents, and what each level of access unlocks.
- Security and data protection What Cyborg7 sends over the network, what it stores and why, how that data is protected, and what it does not do today.
- Troubleshooting Common Cyborg7 issues and how to fix them.
Self-Hosting
- Self-hosting overview What you run when you self-host Cyborg7 (machines, your own relay and database) and where to start.
- Docker (cloud relay) Run the Cyborg7 cloud relay with Docker Compose: relay, Redis and a one-shot migrate job, backed by your own PostgreSQL.
- Production checklist What to check before exposing a self-hosted Cyborg7 relay to your team.