How do I connect Desktop Touch MCP Server to Claude?

Run `claude mcp add desktop-touch-mcp -- npx -y @harusame64/desktop-touch-mcp` in Claude Code, or add the same command and arguments under mcpServers in Cursor's mcp.json or Claude Desktop's claude_desktop_config.json, then restart the client. The blocks above are ready to paste.

Is Desktop Touch MCP Server free?

The listed licence is MIT. Check the upstream terms for permitted use and commercial requirements; a public repository does not by itself mean the software is free or open source. Connected APIs and hosted services may have separate charges.

What can Desktop Touch MCP Server do?

Desktop Touch MCP Server documents 16 tools to the agent, including browser_eval, click_element, desktop_act. The descriptions above come from project documentation. A live handshake does not test individual tool behavior.

Desktop Touch MCP

Unclaimed

by Harusame64

Windows computer-use MCP server: drive any desktop app via semantic discover-then-act targeting (entities + leases, not pixel coordinates), with per-action perception guards, a native Rust UIA engine, Chrome CDP, and Key Locker credential autofill for ssh/sudo password prompts. Works with Claude, Cursor, and any MCP client.

Set up this server

Compare servers My Stack

01 / Choose your client

02 / Before you connect

Authentication is not specified. Check the project instructions before connecting.

Install the runtime required by the project and make it available to your client.

Project instructionsEnvironment variable names (optional)

Use names from the project instructions, separated by commas. Enter names only, never secret values.

03 / Add the configuration

Source: generated from public install instructions

claude mcp add --transport stdio 'desktop-touch-mcp' -- 'npx' '-y' '@harusame64/desktop-touch-mcp'

Run in your terminal after replacing any placeholders.

04 / Check it in your client

Open your client’s MCP settings and confirm the server connects and lists its tools. A copied configuration does not confirm a working connection.

agentanthropicautomationcdpclaudeclaude-codecomputer-usedesktop-automationllmmcpmcp-servermodel-context-protocolnodejsrpascreenshottypescriptui-automationwindows

Is this your MCP server?

This listing was auto-indexed from the public record. Claim it to edit the page, set compatibility and unlock growth tools. Takes under two minutes.

Claim this server

Security profile

Claimed and verified servers get a weekly static scan that shows what the code can reach: external services, environment variables, shell commands, agent configuration folders, plus any dependencies with known advisories. Claim this listing to get one. How the security profile works

16 of 16 tools

Documented tools (16)

From project documentation. A server handshake does not verify each tool’s description or behavior.

browser_eval

Deep inspection via js (scripting), dom (HTML), and appState (SPA data extraction).

click_element

Legacy UIA-based click by name/ID (fallback when entities are unavailable).

desktop_act

Perform actions (click, type, drag, select) on entities via lease validation. Returns semantic diffs — plus an optional roiCapture (changed-region PNG + next-target preview) on visual-only targets.

desktop_discover

Observe the desktop. Returns interactive entities with leases (UIA, CDP, Terminal, Visual SoM).

desktop_state

Lightweight check of focus, active window, cursor, and Auto-Perception attention signal.

excel

Author and run Excel VBA macros via COM. action='runvba' writes a macro into a managed Trusted Location and runs it; action='checkaccessvbom' is a read-only preflight. Runs VBA where formula-only tools cannot. One-time setup: node scripts/enable-access-vbom.mjs.

key_locker

Manage credentials the terminal autofills for you (SSH key passphrases, sudo / login passwords). Secrets are entered once into the locker's own secure dialog and stored encrypted on this machine (Windows DPAPI); they are never shown to the assistant. action='launchconsole' opens an autofill-capable

keyboard

Send keyboard input. Supports background input (WMCHAR) and IME-safe clipboard bypass.

run_macro

Batch up to 50 operations into a single round-trip for maximum efficiency.

screenshot

Multi-mode capture: detail='text' (UIA/OCR), diffMode (P-frame), dotByDot (1:1), and background. Returns a cheap screenshot://by-ref/{id} link to the saved image instead of inlining pixels every time.

scroll

Multi-strategy: raw (notches), toelement, smart (virtual lists), and capture (stitch).

server_status

Diagnostic check for native engine health and feature activation.

terminal

Unified command execution: run (send + wait + read), read (OCR/UIA), and send. run completion modes: quiet, pattern, and exit (waits for the command to finish + returns its exit code — see Terminal command completion).

wait_until

Efficient server-side polling for window, focus, text, or URL state changes.

workspace_launch

Launch apps and auto-detect new HWNDs (supports localized titles).

workspace_snapshot

Instant session orientation: all window thumbnails + UI summaries in one call.

Tool change history

Compared across complete checks of the same configuration. Tools were listed, not invoked. Input-schema changes are not measured here.

No complete tool checks yet.

FAQ

Questions about Desktop Touch MCP Server

How do I connect Desktop Touch MCP Server to Claude?: Run `claude mcp add desktop-touch-mcp -- npx -y @harusame64/desktop-touch-mcp` in Claude Code, or add the same command and arguments under mcpServers in Cursor's mcp.json or Claude Desktop's claude_desktop_config.json, then restart the client. The blocks above are ready to paste.
Is Desktop Touch MCP Server free?: The listed licence is MIT. Check the upstream terms for permitted use and commercial requirements; a public repository does not by itself mean the software is free or open source. Connected APIs and hosted services may have separate charges.
What can Desktop Touch MCP Server do?: Desktop Touch MCP Server documents 16 tools to the agent, including browser_eval, click_element, desktop_act. The descriptions above come from project documentation. A live handshake does not test individual tool behavior.