How to Use PigeonCast MCP

1. Record the MCP path. For installations from the official website, the default path is C:\Program Files\PigeonCast\Tools\pigeoncast-mcp.exe; for the Microsoft Store version, it is the bundled Tools\pigeoncast-mcp.exe inside the app package. You can search for this filename in the installation directory. a. If you checked Add to PATH during installation from the official website, you can run pigeoncast-cli ping directly in the terminal. For MCP configuration, it is still recommended to use the full path to pigeoncast-mcp.exe so the AI client can always find the program.
2. Change the command value in the JSON below to the actual path of pigeoncast-mcp.exe on your machine. Note that backslashes in JSON must be written as \. Copy the JSON with the corrected path into the AI chat window and tell the AI: Apply to Global.
3. After saving the configuration, reload MCP in the relevant client or restart the application. You should then see tools whose names start with host_ in the tool list, such as host_ping, host_discovery_list_devices, and host_mirror_start.
4. Quick self-check: tell the AI, "Use PigeonCast's host_ping to check whether the application is running." If the main program is already running, it should return success; if not, it will report that the pipe connection failed.

Option A: Current Project Only (Recommended for Team Sharing)

1. Create .cursor/mcp.json in the project root.
2. Paste in the mcpServers configuration from the previous section, with command set to the absolute path of pigeoncast-mcp.exe on the local machine.
3. Open Cursor → Settings → MCP, confirm that the pigeoncast server appears, then click Reload or refresh.

Option B: Available to All Projects

1. Go to Cursor → Settings → MCP → Add new global MCP server.
2. Select stdio as the type, enter the full path of pigeoncast-mcp.exe as the command, and leave the arguments empty.Alternatively, manually edit the user-level configuration file, which varies by version and is commonly %USERPROFILE%.cursor\mcp.json, using the same format as the previous section.

Tips

1. Make sure the PigeonCast main application is running before asking the agent to search for devices or start screen casting.
2. If Cursor is still holding on to an old MCP process, reload after changing the configuration. If problems remain, fully quit and reopen Cursor.
3. For detailed tool descriptions and testing scenarios, see mcp-usage-for-testers.md in the repository.

Option A: Add from the Command Line (Recommended)

Run the following in PowerShell, replacing the path with your installation path:claude mcp add --scope user pigeoncast -- "C:\Program Files\PigeonCast\Tools\pigeoncast-mcp.exe" If you want it available only for the current project:

claude mcp add --scope project pigeoncast -- "C:\Program Files\PigeonCast\Tools\pigeoncast-mcp.exe"

To verify that it was added successfully:

claude mcp list

Option B: Edit the Configuration File

Example .mcp.json:

{
  "mcpServers": {
    "pigeoncast": {
      "type": "stdio",
      "command": "C:\Program Files\PigeonCast\Tools\pigeoncast-mcp.exe",
      "args": []
    }
  }
}

After saving, restart the Claude Code session, or run the MCP reload command inside the session, depending on the current version.

1. Fully quit Claude Desktop, including the system tray process.
2. Edit the configuration file. If you are using the MSIX installation, use the Actually read by the app path below instead of only editing the file opened by the Edit Config button, or the change may not take effect.
3. At the root level of the JSON file, add or merge mcpServers using the same structure as Section 2, with command set to the absolute path of pigeoncast-mcp.exe.
4. Save the file and restart Claude Desktop.
5. In Settings → Developer, confirm that pigeoncast appears in the MCP server list.

Codex uses TOML to configure MCP. The default file is %USERPROFILE%.codex\config.toml. At the project level, you can use .codex/config.toml in a trusted repository root.

Option A: Add from the Command Line

codex mcp add pigeoncast -- "C:\Program Files\PigeonCast\Tools\pigeoncast-mcp.exe"

To view the current configuration:

codex mcp list

Option B: Edit config.toml Add the following to ~/.codex/config.toml:

[mcp_servers.pigeoncast]
command = "C:\Program Files\PigeonCast\Tools\pigeoncast-mcp.exe"
args = []
enabled = true

If you use a project-level .codex/config.toml, you must mark that project as Trusted, or the local MCP server may not load. Documentation: Codex MCP Servers (follow the current OpenAI docs when in doubt)

Any client that supports stdio-based MCP and lets you specify an executable path plus arguments can use the same basic configuration:

• command: the absolute path to pigeoncast-mcp.exe
• args: [] (leave empty)
• env: usually no extra environment variables are required Take the mcpServers block from Section 2 and adapt the filename and field names to match the target client. Some clients use fields such as type: "stdio".