WpfPilot MCP
MCP server for AI agents to automate Windows WPF apps with semantic UI Automation selectors, screenshots, recording, and FlaUI test generation.
README
WpfPilot MCP
WpfPilot MCP is a local Model Context Protocol server for Windows WPF applications. It lets AI coding agents inspect UI Automation trees, click and type through semantic selectors, diagnose WPF-specific issues, record workflows, and generate xUnit + FlaUI tests.
Features
- Attach to or launch WPF processes.
- Capture semantic UI snapshots (selectors, not screen coordinates).
- Query text, value, state, bounds, patterns, children, ancestors, siblings, and selection.
- Act with verbs such as click, set value, select, toggle, expand, collapse, scroll, and drag/drop.
- Wait and assert on UI state with structured errors.
- Capture screenshots.
- Record workflows and generate test code.
- Optional in-process probe for ViewModel, binding, command, validation, and dispatcher diagnostics.
Requirements
- Windows 10/11.
- A WPF application to automate.
- Node.js 18+ if using
npx(recommended). No .NET SDK required for the release binary or npm launcher.
Getting started
Add this to your MCP client configuration (global or project-scoped):
{
"mcpServers": {
"wpfpilot-mcp": {
"type": "stdio",
"command": "npx",
"args": ["-y", "@skuzadev/wpfpilot-mcp"]
}
}
}
Or run directly (stdio; waits for an MCP client):
npx -y @skuzadev/wpfpilot-mcp
The npm launcher downloads the latest Windows release binary on first run, then proxies stdio to it.
Other installs (persistent wpfpilot-mcp command, release zip, uninstall): see Install WpfPilot.
Popular clients
Use the standard config above unless noted. More clients: docs/all-clients.md.
<details> <summary>Cursor</summary>
Global: ~/.cursor/mcp.json (do not add a project .cursor/mcp.json unless you need a repo-specific override)
{
"mcpServers": {
"wpfpilot-mcp": {
"type": "stdio",
"command": "npx",
"args": ["-y", "@skuzadev/wpfpilot-mcp"]
}
}
}
Or: Cursor Settings → MCP → Add new MCP Server — command npx, args -y @skuzadev/wpfpilot-mcp.
cursor-agent mcp list
cursor-agent mcp list-tools wpfpilot-mcp
</details>
<details> <summary>Codex</summary>
codex mcp add wpfpilot-mcp -- npx -y @skuzadev/wpfpilot-mcp
codex mcp list
Or ~/.codex/config.toml:
[mcp_servers.wpfpilot-mcp]
command = "npx"
args = ["-y", "@skuzadev/wpfpilot-mcp"]
enabled = true
startup_timeout_sec = 30
tool_timeout_sec = 60
</details>
<details> <summary>Claude Code</summary>
claude mcp add --transport stdio wpfpilot-mcp -- npx -y @skuzadev/wpfpilot-mcp
claude mcp list
Project .mcp.json:
{
"mcpServers": {
"wpfpilot-mcp": {
"type": "stdio",
"command": "npx",
"args": ["-y", "@skuzadev/wpfpilot-mcp"]
}
}
}
</details>
<details> <summary>Claude Desktop</summary>
Windows config: %APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"wpfpilot-mcp": {
"type": "stdio",
"command": "npx",
"args": ["-y", "@skuzadev/wpfpilot-mcp"]
}
}
}
Restart Claude Desktop after saving.
</details>
<details> <summary>VS Code</summary>
Create .vscode/mcp.json:
{
"servers": {
"wpfpilot-mcp": {
"type": "stdio",
"command": "npx",
"args": ["-y", "@skuzadev/wpfpilot-mcp"]
}
}
}
Or: code --add-mcp '{"name":"wpfpilot-mcp","command":"npx","args":["-y","@skuzadev/wpfpilot-mcp"]}'
</details>
First prompt
After connecting the MCP client, ask:
List the WpfPilot tools and attach to my running WPF application.
Useful follow-up prompts:
Show the main window UI tree.
Click the Save button using a selector, not coordinates.
Wait until the status text says Saved.
Why is the Submit button disabled?
Record this workflow and generate an xUnit test.
Tools
| Tool | Purpose |
|---|---|
wpf_capabilities |
List supported verbs, query kinds, and wait conditions. |
wpf_query |
Read UI state. |
wpf_act |
Perform UI actions. |
wpf_wait |
Wait for UI state. |
wpf_assert |
Verify UI state. |
Full tool list: docs/tools-reference.md.
Optional WPF probe
The probe runs inside your WPF process and exposes diagnostics that UI Automation cannot see directly (bindings, ViewModels, commands, validation).
Use wpf_probe_connect, then inspect my ViewModel and binding errors.
Setup: docs/probe-setup.md.
Safety
WpfPilot is intended for local development and test automation.
- It runs as your user account and can interact with UI visible to that account.
- It does not expose general shell, registry, or arbitrary filesystem tools through MCP.
- Mutating UI actions are audited under the user's local app data folder.
- The probe requires explicit installation in the target WPF app.
- Treat every MCP server as trusted local code before enabling it in an agent.
Troubleshooting
wpfpilot-mcp is not recognized
Restart your terminal after running the installer, or use the full path to wpfpilot-mcp.exe in your MCP client configuration. See all-clients.md.
Server starts but no tools appear
Restart the MCP client and check its MCP logs. Also verify npx -y @skuzadev/wpfpilot-mcp runs from a normal terminal.
Cannot attach to an app
Make sure the WPF app is running in the same user session and at a compatible privilege level. If the app runs as administrator, the MCP client may also need to run elevated.
Probe cannot connect
Confirm the target app called ProbeHost.Start(), then use wpf_probe_status and wpf_probe_connect. If needed, pass wpfpilot-mcp-probe-{ProcessId} explicitly.
Documentation
Contributing
See CONTRIBUTING.md.
License
MIT. See LICENSE.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.