Hey ​@Jiyun! I checked in with the engineering team that works on MCP development/improvements, and I wanted to check on a couple of things. 

Are you (and the rest of the team you’re working with) using the desktop app exclusively? If so, can you verify you’re on the most recent version of the desktop app (125.5.6 -- this is mac)? Does this travel across both app and the Figma website?

If you regenerate the code, does it still not render as expected?

Let me know when you have a sec.

Hi, ​@ksn! we initially understood #get_image as a tool to retrieve the actual image used in the design. However, upon further review, we realized that it’s intended to provide visual context to the LLM through a screenshot. It seems we misunderstood the purpose of the tool and used it incorrectly.

Thanks ​@Jiyun! No worries - MCP is still new, and having this conversation is welcomed. It can help us determine if we need to provide more clarification in any of our published documentation (help center, etc.)

I’m going to close out this topic, but feel free to create a new one if any other questions come up