Omni Computer Use — Development skill for Claude Code

omni-computer-use

A faithful **1:1 replica of Anthropic's official `computer-use` tool surface** — shipped as a plain **MCP server** for Windows. It lets the **Claude Code CLI** (or any MCP client) drive the Windows desktop the way Claude Desktop's built-in computer-use does: screenshots, mouse / keyboard / scroll / drag / batch input, clipboard, multi-monitor, and application launch.


*Real session, one take: the CLI activates omni (glow + pill + terminal parked to the corner), opens VS Code, and types. The corner terminal is the actual conversation that produced this GIF.*

Same tool names, same parameters, same coordinate semantics as the desktop tool — an agent that already knows Anthropic's computer-use works here unchanged. Built and verified against Claude Desktop's own `computer-use` as the ground-truth oracle.

Vision-and-coordinate based, like the official tool (rather than a UI-tree / accessibility approach) — for when you want Anthropic's computer-use paradigm on a Windows CLI: desktop automation, GUI testing, or letting the agent verify the app you're building.

Install

Requires Windows and Python 3.11+ (uv / uvx fetch Python for you). The PyPI package name carries an `-mcp` suffix; everything else — the command, the server key, this repo — is plain `omni-computer-use`.

**Recommended — install into Claude Code.** Claude Desktop reads Claude Code's MCP servers *in addition to* its own, so this one command makes the server available in both the Claude Code CLI and the Claude Desktop app:

claude mcp add omni-computer-use -s user -- uvx omni-computer-use-mcp
claude mcp list   # expect: omni-computer-use … ✓ Connected

The other directions are narrower. In Claude Desktop's `claude_desktop_config.json` — visible to Claude Desktop only, the CLI won't see it: