Welcome to CrossMind

CrossMind is an agent workspace: you create AI agents, point them at work (growth, research, operations), and they run sessions on your schedule, locally on your own computers or in the cloud.

What CrossMind does

• Sessions: chat with an agent (Copilot) or hand it a bounded task it runs start-to-finish on its own. Streaming output, tool approvals, context metering, and per-message model switching are built in.
• Scheduled tasks: run any prompt on a cron or interval, with catch-up after downtime.
• Skills: reusable playbooks an agent follows; install from a URL, a Git repo, or share them publicly.
• MCP tools: extend agents with Model Context Protocol servers; see the token cost of every tool before you enable it.
• Computers: pair your laptop or desktop as a runtime host. Agents then run with your local CLIs (Claude Code, Codex, Cursor, ZCode and more) and your files.
• Integrations: connect X, LinkedIn, GitHub, Notion, Telegram, Stripe, Vercel and more; agents get the credentials injected automatically.

Where to go next

1. Core concepts: Agents, sessions, runtimes and computers in two minutes.
2. Getting started: create your account, first agent, and first session.
3. Browse the sidebar for any feature in depth.

Mobile, desktop & iPad

CrossMind ships a macOS desktop app and iOS/iPadOS apps alongside the web app. Sessions, approvals and notifications follow you across devices. Start a run on the web, review the results on your iPad, approve its tool calls from your phone.

[Screenshot pending: iPad landscape (native app, device bezel): copilot session with tool cards — /docs-media/en/ipad-session.png]

iPhone (native app) screenshots — pending capture

• An active copilot session with tool cards (ios-session.png)
• The new-session composer (ios-new-run.png)
• An inline approval card (ios-approval.png)

[Screenshot pending: iPad landscape (native app, device bezel): memos board with lifecycle columns — /docs-media/en/ipad-memos.png]