[FEATURE] claude-in-chrome: add filePath param to computer tool screenshot action

Preflight Checklist

• [x] I have searched existing requests and this feature hasn't been requested yet
• [x] This is a single feature request (not multiple features)

Problem Statement

The mcp__claude-in-chrome__computer tool's screenshot action captures screenshots and returns an in-memory imageId (e.g., ss_xxx), but there is no way to save that screenshot to a file on disk.

Claude-in-Chrome is the only MCP browser tool that uses the user's real, logged-in Chrome session — essential for screenshotting authenticated pages (ChatGPT, Gemini, ElevenLabs, etc.). But without a save-to-disk option, the screenshots are trapped in memory with no way to persist them as project assets.

The separate chrome-devtools-mcp server has a filePath parameter on its take_screenshot tool, but it runs a different browser instance that is not logged in anywhere — so it can't capture authenticated pages.

Proposed Solution

Add an optional filePath parameter to the computer tool's screenshot action, identical to how chrome-devtools-mcp implements it:

mcp__claude-in-chrome__computer
  action: screenshot
  tabId: 12345
  filePath: "/path/to/project/assets/screenshot.png"

When filePath is provided, save the screenshot to that path on disk (in addition to returning the imageId). When omitted, behavior stays the same as today.

Alternative Solutions

Workarounds tried — all fragile or broken:

1. chrome-devtools take_screenshot with filePath — works for saving to disk, but runs a separate browser that is not logged in to any service
2. html2canvas via javascript_tool — breaks on cross-origin resources and CSP restrictions on most sites
3. gif_creator with download — wrong format (GIF not PNG), includes visual overlays (click indicators, watermark)
4. upload_image — uploads imageId to web form inputs, cannot save to local filesystem

None of these provide a reliable path from "logged-in browser page → file on disk."

Priority

Critical - Blocking my work

Feature Category

MCP server integration

Use Case Example

1. I'm writing an ebook about AI tools with screenshots of ChatGPT, Gemini, ElevenLabs, HeyGen, Kling
2. I need to capture UI screenshots of these tools and save them as PNG files in ebook-assets/
3. These pages require authentication — only claude-in-chrome can access them (logged-in session)
4. Currently I can SEE the screenshots in claude-in-chrome but cannot SAVE them to disk
5. With filePath, I could: navigate → screenshot with filePath → done. One tool call instead of a broken workaround chain.

Additional Context

• chrome-devtools-mcp already implements this exact pattern via its take_screenshot tool with filePath — proving the concept works
• The native messaging host already has filesystem access (it reads/writes files for other operations), so adding file-write for screenshots should be straightforward
• Reference: chrome-devtools-mcp PR #172 added filePath to their take_screenshot