From 33b93296bdb64ab5033400f2f5424c064d6f0152 Mon Sep 17 00:00:00 2001
From: Maruen Mehana <maruen@gmail.com>
Date: Wed, 29 Apr 2026 14:04:28 -0300
Subject: [PATCH] feat: add Claude plugin executablePath user config
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Add a userConfig option for the executablePath CLI flag so users running
the plugin via Claude Code can point at a custom Chrome binary without
editing the plugin manifest in their plugin cache. Useful when the host
has multiple Chrome installations — for example pointing at Chrome for
Testing on macOS so the MCP-launched Chrome has a separate bundle ID
(com.google.chrome.for.testing) and can be scripted unambiguously by
AppleScript / OS automation tools without conflicting with the user's
regular Chrome (com.google.Chrome).

Mirrors the user_config.browserUrl pattern from #1851.

Leaving the option blank keeps the existing auto-detected-stable-Chrome
behavior — no opt-in cost.
---
 .claude-plugin/plugin.json |  9 ++++++++-
 README.md                  | 16 ++++++++++++++++
 2 files changed, 24 insertions(+), 1 deletion(-)

diff --git a/.claude-plugin/plugin.json b/.claude-plugin/plugin.json
index 7bbde5828..2fc6cfa06 100644
--- a/.claude-plugin/plugin.json
+++ b/.claude-plugin/plugin.json
@@ -2,11 +2,18 @@
   "name": "chrome-devtools-mcp",
   "version": "0.23.0",
   "description": "Reliable automation, in-depth debugging, and performance analysis in Chrome using Chrome DevTools and Puppeteer",
+  "userConfig": {
+    "executablePath": {
+      "description": "Optional path to a custom Chrome executable. Useful when the system has multiple Chrome installations (e.g. point at the Chrome for Testing binary so the MCP-launched Chrome stays separate from the user's regular Chrome and can be scripted unambiguously by AppleScript / OS automation tools). Leave blank to use the auto-detected stable Chrome.",
+      "sensitive": false
+    }
+  },
   "mcpServers": {
     "chrome-devtools": {
       "command": "npx",
       "args": [
-        "chrome-devtools-mcp@latest"
+        "chrome-devtools-mcp@latest",
+        "--executablePath=${user_config.executablePath}"
       ]
     }
   }
diff --git a/README.md b/README.md
index 9d18f1ab5..5c372a102 100644
--- a/README.md
+++ b/README.md
@@ -166,6 +166,22 @@ Then, install the plugin:
 
 Restart Claude Code to have the MCP server and skills load (check with `/skills`).
 
+To point the plugin at a custom Chrome binary (e.g. [Chrome for Testing](https://developer.chrome.com/blog/chrome-for-testing/) so the MCP-launched Chrome has a separate bundle ID from the user's regular Chrome and can be scripted unambiguously by macOS AppleScript), set the plugin's `executablePath` option in your Claude Code settings:
+
+```json
+{
+  "pluginConfigs": {
+    "chrome-devtools-mcp@claude-plugins-official": {
+      "options": {
+        "executablePath": "/Applications/Google Chrome for Testing.app/Contents/MacOS/Google Chrome for Testing"
+      }
+    }
+  }
+}
+```
+
+Leave `executablePath` blank to use the default behavior where Chrome DevTools MCP auto-detects stable Chrome.
+
 > [!TIP]
 > If the plugin installation fails with a `Failed to clone repository` error (e.g., HTTPS connectivity issues behind a corporate firewall), see the [troubleshooting guide](./docs/troubleshooting.md#claude-code-plugin-installation-fails-with-failed-to-clone-repository) for workarounds, or use the CLI installation method above instead.