Harusame64/desktop-touch-mcp
by Various
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 nat
MCP
Harusame64/desktop-touch-mcp
Added 1 June 2026
Overview
A Windows computer-use MCP server that drives any desktop app via semantic discover-then-act targeting using entities and leases instead of pixel coordinates. It combines a native Rust UI Automation engine with Chrome DevTools Protocol integration and includes per-action perception guards for safer automation.
Best for
Best for
Developers building Windows desktop automation workflows that need a semantic, reliable alternative to traditional screen-scraping.
Use cases
- Automate repetitive GUI tasks in Windows applications using natural language from an MCP client.
- Perform semantic UI navigation in legacy or non-accessible apps with the Rust UIA engine.
- Combine desktop and browser automation in a single MCP session for cross-application workflows.
Notes
A Windows computer-use MCP server that drives any desktop app via semantic discover-then-act targeting using entities and leases instead of pixel coordinates. It combines a native Rust UI Automation engine with Chrome DevTools Protocol integration and includes per-action perception guards for safer automation.
0 stars on GitHub. Last updated 2026-05-25. Licensed MIT.
Use cases
- Automate repetitive GUI tasks in Windows applications using natural language from an MCP client.
- Perform semantic UI navigation in legacy or non-accessible apps with the Rust UIA engine.
- Combine desktop and browser automation in a single MCP session for cross-application workflows.
Pros
- Semantic targeting avoids brittle pixel coordinates, improving automation reliability.
- Per-action perception guards reduce the risk of unintended operations.
- Supports both native Windows apps and Chrome via CDP for broad desktop control.
Cons
- Windows-only with no cross-platform support.
- New project with zero stars and limited community adoption, so long-term stability is unproven.
- Requires understanding the lease-based targeting model and setting up an MCP client.
Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.
Pros
- Semantic targeting avoids brittle pixel coordinates, improving automation reliability.
- Per-action perception guards reduce the risk of unintended operations.
- Supports both native Windows apps and Chrome via CDP for broad desktop control.
Cons
- Windows-only with no cross-platform support.
- New project with zero stars and limited community adoption, so long-term stability is unproven.
- Requires understanding the lease-based targeting model and setting up an MCP client.
Pairs with
Other entries in the index that connect to this one. Click through to see the chain.
Get the free Developer’s Field Guide
A 27-page field guide to the AI coding workflow with Claude. Claude Code, MCP servers, the prompt patterns that work, and what to delegate. Free.
Enter your work email. We send it straight over, plus a few short notes worth knowing. Unsubscribe any time.