Capsulate MCPサーバー

@capsulate/mcp

The Capsulate Model Context Protocol server. Give Claude Code, Claude Desktop, Cursor, Windsurf, Antigravity, any MCP client, the ability to launch and drive isolated cloud computers ("capsules"), including real Linux and Windows desktops that a person can watch live and take over.

Setup

1. Create an API key in the Capsulate dashboard: Settings → API & Keys.
2. Add the server to your MCP client config.

Claude Code

claude mcp add capsulate -e CAPSULATE_API_KEY=cap_live_... -- npx -y @capsulate/mcp

Claude Desktop / Cursor

{
  "mcpServers": {
    "capsulate": {
      "command": "npx",
      "args": ["-y", "@capsulate/mcp"],
      "env": {
        "CAPSULATE_API_KEY": "cap_live_…"
      }
    }
  }
}

To keep the key out of the config file, put it in a file and point at it with CAPSULATE_API_KEY_FILE instead. The file holds either the bare key or a CAPSULATE_API_KEY=cap_live_... line.

For a self-hosted or non-default deployment, also set "CAPSULATE_BASE_URL": "https://api.your-domain".

Tools

Tool	What it does
capsulate_launch	Launch a capsule and wait until ready
capsulate_list / capsulate_get	Inspect capsules
capsulate_destroy	Tear a capsule down
capsulate_exec	Run a command (argv array)
capsulate_run_code	Run a python/node/bash snippet
capsulate_write_file / capsulate_read_file	File I/O
capsulate_set_env	Set environment variables
capsulate_snapshot	Snapshot a capsule
capsulate_catalog	Discover launchable templates & bundles

Desktop computer use

Tool	What it does
capsulate_screenshot	See the screen (returns an image; format: "jpeg", a scale or a region make it smaller)
capsulate_screen_info	The screen's size, the pointer and the active window
capsulate_click / capsulate_move / capsulate_drag / capsulate_scroll	Pointer
capsulate_type / capsulate_key	Keyboard
capsulate_ui_tree	Read the accessibility tree (AT-SPI on Linux, UI Automation on Windows) with clickable centers
capsulate_find_element / capsulate_wait_element	Find or wait for an element by role, name or text
capsulate_click_element	Find an element and click it: the reliable way to press a named button
capsulate_clipboard	Read or set the clipboard (paste long text instead of typing it)
capsulate_element	Act on an element without coordinates: set its value, clear, toggle, expand or collapse, invoke, select an option by name, focus, scroll it into view
capsulate_windows	List windows or bring one to the front
capsulate_window	Minimize, maximize, restore, close, move or resize a window
capsulate_open	Open a URL, a file or an app by name, and wait for its window
capsulate_wait_for	Wait for an element to appear, go away, be enabled or show a text, for a window, or for the screen to settle
capsulate_ocr	Read the text on screen when the UI tree has no labels
capsulate_run_actions	Several steps in one call, back to back, stopping at the first that fails, with a look at the end

The input verbs accept screenshot: true (and screenshotDelayMs) to return the frame after the action. Every tool works the same on Linux and Windows desktops; a multi-line editor is a document on Windows and a text on Linux.

Browser fast path

The capsule's own Chrome, driven through the page's structure at page speed, on the same screen the person watches: capsulate_browser_open, _navigate, _tabs, _snapshot (an outline with refs), _click, _fill (the value is never echoed), _select, _press, _scroll, _wait, _text, _eval and _screenshot.

Live: a person watches, takes over, hands back

A desktop capsule has one shared screen. The AI drives it through the tools above while a person watches the same screen in a browser, sees the AI's pointer, clicks and keys drawn over it, reads its narration, and can take the screen at any time.

Tool	What it does
capsulate_console_link	A signed, expiring URL that opens the live screen in any browser with no login (control or view)
capsulate_say	Show one line of narration on the live console
capsulate_control	Who is driving: free, agent (the AI) or human; take or release the screen; set the pace (natural or presentation) for a person watching
capsulate_handoff	Hand the screen to the person with a note ("Please sign in"), wait until they hand it back, get their note
capsulate_wait_for_handback	Keep waiting for the person to hand the screen back

While a person holds the screen, the AI's input verbs answer human_in_control and nothing reaches the capsule; the tool result says to wait with capsulate_wait_for_handback. Reading the screen (screenshot, UI tree, OCR) keeps working. Each AI input holds the screen for 45 seconds; the console is view-only for everyone else during that time and turns interactive again when the AI goes quiet or releases.

Skills: learn once, replay in minutes

Tool	What it does
capsulate_lesson	Learn a job on a capsule: a person teaches on the shared screen (clicks, keys and typing with the elements they hit, passwords masked, optionally video and voice) or the AI teaches itself toward a goal; get reads the digest
capsulate_skills	The team's skills: save one from a lesson's digest, versions, approve, save an environment, rehearse, weekly rehearsals
capsulate_skill_run	Run a skill's procedure in the background; it asks you where the skill leaves a decision open
capsulate_inbox	Runs that wait, versions to approve, failed rehearsals and skills whose software changed

The server's instructions carry the learning protocol: explore with a goal, note what you learn, finish, read the digest, save the skill; next time launch its environment and run its procedure.

Try it

"Spin up an ubuntu-22.04-headless capsule, run my test suite, and show me the output."

The model calls capsulate_launch, then capsulate_exec, and reports back.

"Launch an ubuntu-22.04-desktop capsule, give me a live link, and find the latest Firefox release notes. If anything needs a login, hand it to me."

The model launches the desktop, calls capsulate_console_link and gives you the URL, narrates with capsulate_say as it works, and calls capsulate_handoff when it needs you. You press Done, hand back in the live console when you finish, and it carries on.

Develop

npm run build                          # from the repo root: sdk, ai-tools, cli, mcp
node apps/mcp/scripts/smoke-live.mjs   # live tools end to end against a fake API