Using IconVectors with VS Code Copilot

VS Code's MCP support lets GitHub Copilot talk directly to IconVectors. Copilot can inspect the current drawing, create new geometry, edit paths, render previews, and export assets by calling the MCP tools exposed by the running editor.

Current public setup note: the example paths below use the Windows 1.40 release. The macOS MCP package is being prepared.

1. Start IconVectors

& 'C:\Program Files\Axialis\IconVectors\IconVectors.exe'

2. Create the MCP configuration

Use either a workspace file .vscode/mcp.json or your VS Code user MCP configuration and add:

{
  "servers": {
    "iconvectors": {
      "type": "stdio",
      "command": "C:\\Program Files\\Axialis\\IconVectors\\IconVectorsMcp.exe",
      "args": ["--port", "61337"]
    }
  }
}

If you need to force the working directory, wrap the command and change to C:\Program Files\Axialis\IconVectors before launching the sidecar.

3. Verify in Copilot Chat

• Make sure the server appears in VS Code's MCP UI.
• Switch Copilot Chat to Agent mode.
• Try prompts such as Use IconVectors to call app.ping. and Create a new 32x32 document named test icon and return SwiftUI code for it.

Helper files

Use Help -> Open MCP Files Folder to open the installed helper bundle. The files most relevant to this host are vscode-copilot-implementation-instructions.md, iconvectors-mcp-command-reference.md, and vscode-copilot-AGENTS.md.template.