Add a Puppeteer tool to capture screenshot as raw base64 text

[kim00jun]

Description

Added an optional param(encoded) to puppeteer_screenshot tool to capture screenshots and return the result as a raw base64-encoded string (without a MIME type).

Server Details

• Server: Puppeteer
• Changes to: tools (puppeteer_screenshot)

Motivation and Context

The existing puppeteer_screenshot tool returns screenshots as a mimeType: "image/png". However, some MCP clients are unable to properly parse or display image type content.

This optional param solves that by returning only the raw base64 string of the screenshot, enabling better compatibility with clients that expect raw text output.

How Has This Been Tested?

Tested locally with an LLM client using:

• Full page screenshots
• Element-specific screenshots
• Verified that:
  • base64 strings are returned without MIME prefix
  • raw encoded image texts are correctly recognized by MPC client
  • no errors occurred during Puppeteer execution

Breaking Changes

No breaking changes. This is an additive feature.

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)
• Documentation update

Checklist

• I have read the MCP Protocol Documentation
• My changes follows MCP security best practices
• I have updated the server's README accordingly
• I have tested this with an LLM client
• My code follows the repository's style guidelines
• New and existing tests pass locally
• I have added appropriate error handling
• I have documented all environment variables and configuration options

Additional context

[olaservo]

Hi @yomandawg sorry for missing your tagging me here. I added a comment and will keep an eye out for updates on this one. Thanks!

[olaservo]

Looks good, just wanted to double check that this will do what you originally intended since in the original PR you said:

This optional param solves that by returning only the raw base64 string of the screenshot, enabling better compatibility with clients that expect raw text output.

Technically if encoded is true you are also including the data:image/png;base64 here and not just the raw string. The actual implementation you added makes sense here though if you're trying to work with hosts that only work with TextContent types, and/or if you want to embed the result directly in an img src.

[kim00jun]

Yes, the updated PR message accurately reflects the commit. And you're also right-- although image data will be included in both cases, correctly identifying it as either TextContent or ImageContent is essential for proper differentiation. That's where the optional param comes into play.
Thanks for the review!

[olaservo]

(Fyi, this will go to an package publishing approval workflow and get tagged and pushed to NPM as a new version once that gets approved by someone on the core team.)