[Feature Request] Claude Chrome extension screenshots should support local file saving like Playwright MCP

[tfayyaz]

Preflight Checklist

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

Problem Statement

Bug Description
You cannot access screenshots taken by claude chrome to embed in a markdown document. It would be useful if you could choose to save screenshots locally as you can with Playwright MCP

I asked Claude Code and it said:

Correct. The screenshots taken with the Claude Chrome extension
(mcp__claude-in-chrome__computer with action "screenshot") are:

1. Displayed in the conversation - I can see them when they're returned
2. Given reference IDs (like ss_92495b3gs) - for use with tools like upload_image
3. NOT saved to disk - they're not written as files to the filesystem

So I cannot:

• Access them later as files
• Convert them to base64
• Embed them in a markdown document

Environment Info

• Platform: darwin
• Terminal: vscode
• Version: 2.0.73
• Feedback ID: 542516aa-dc09-47ba-8567-301b434a9901

Proposed Solution

Claude should save screenshots to a local screenshots folder in the cwd

Alternative Solutions

No response

Priority

Medium - Would be very helpful

Feature Category

Other

Use Case Example

No response

Additional Context

No response

[github-actions]

This issue has been inactive for 30 days. If the issue is still occurring, please comment to let us know. Otherwise, this issue will be automatically closed in 30 days for housekeeping purposes.

[rstrand]

This issue has been inactive for 30 days. If the issue is still occurring, please comment to let us know. Otherwise, this issue will be automatically closed in 30 days for housekeeping purposes.

This still isn't implemented and would be useful for me too

[varun-d]

Yea, I am curious what happens to the claude code chrome plugin screenshots. Where are they saved and are these used for training later even on Pro / Max accounts?

[prakersh]

If what you need is structured data about the page rather than screenshots, onUI might help. It's a free, open-source browser extension with a local MCP server — AI agents can query annotations (element labels, coordinates, metadata) directly, which are already structured text data rather than image files.