Ask the Community

Hi everyone,I’m running into an issue with Auto Layout and component variants and I’m not sure what the correct approach is.I have a button component with two variants:- Default- Hold / While pressing, where the button slightly scales up (as a press feedback)When the component is used on its own, the scale difference between Default and Hold works as expected.However, when I place the button inside a mobile screen frame with Auto Layout, I need to set the button to Width = Fill container to make it responsive. The problem is: in this case, when I switch to the Hold variant, the scale effect is no longer visible: Default and Hold look identical, because Auto Layout keeps the width constrained to the parent.So my questions are:Is this a known limitation of Auto Layout with scaling variants?Is there a recommended way to represent a Hold state that involves scaling when the component must use Fill?From a handoff perspective, should this kind of interaction be documented rather than visuall

Asking because my dev team is having trouble locating exact screens they need because the keywords they are searching include items and strings that are cropped out/not visible in the frame. The component is not hidden, just cropped out. Is it possible to search ONLY visible elements in a frame? This would exclude components hidden behind larger ones and also cropped components. I tried searching for plugins and didn’t see this type of question surface yet. I get that it’s probably a super niche case so any kind of answers would be appreciated

Hello guys! I’m building a plugin for my design team that uses the isAsset() function. In one part, my plugin exams all nested children within a given node and returns whether each children is an asset.The function is doing a pretty good job recognizing icons but I ran into some issues with nested components. For example in the screenshot below, the isAsset() function returns true for two layers within the Instagram component, but only one within the facebook component. There were also a few times that running the function on the same component returns different results. Does anyone have similar experience? Is there a workaround to this?  

I created a file and sent out a link for dev on Nov. 30th. Since then I have made updates to the file. Suddenly, the link I provide comes up empty and the file is missing. However, there are 3 copies. One is more of a starter file and the other two only show a version history of Nov. 30th and neither have the recent updates. I did not delete the original file nor make the copies, at least not that I’m aware of. I guess i’m going to edit one of the copies and go from there but would love to know how this happened. Anyone have any clue?

I'm currently experimenting with our design system library file. Whenever I select a library in Figma, the generated global CSS changes. However, when I add our library, the generated global CSS doesn't reflect our color variables, typography styles, or effects defined in Figma.How can I generate a proper global CSS that accurately uses the color variables, font styles, and effects from our actual Figma library?

I followed this doc https://developers.figma.com/docs/figma-mcp-server/local-server-installation/ to enable MCP server. Also, setup claude code as MCP Client. However, when I prompt “Implement this Figma design. {DESIGN_URL}” claude code tries to fetch design details but fails with error code 403.I see mcp server configured correctly. 

Hello — I’m a developer trying to get my website online. Everything goes smoothly until the final stretch of publishing. We’re using Figma make with AI to design it and code it. Around 90% into the deploy, the publish process suddenly fails and I’m hit with a slew of CSS errors. I know the AI assistant attempted to upgrade the project to CSS Level 4, and I reverted it back to CSS Level 3 for compatibility/support reasons. However, when I try to deploy, I’m still seeing CSS errors. Despite multiple checks, the AI keeps suggesting this may be an issue on Figma’s side. The site is alignment-ai.io and it’s currently deployed, if you’d like to inspect it with DevTools to see the errors directly. I’ve tried everything I can control on my end: cache busting, dozens of code checks, removing our Google Analytics tag, and rolling back to prior versions. No luck — the site still goes offline. This also happened on a previous build around New Year’s, so we duplicated the project into a fresh build

Hello,I'm trying to embed a figma-prototype into our application where I need to use the Embed API to track the movement of respondents through the prototype.I need to read PRESENTED_NODE_CHANGED and INITIAL_LOAD.I need to set last visited node after LOAD.I'm having trouble getting the Embed API to work for unlogged users (no figma session, no figma cookie).Your example on the website do that. And according to the documentation it should be possible.https://developers.figma.com/docs/embeds/embed-api/#example-index-filePrototype URLhttps://embed.figma.com/proto/W6MiVVVuvJb5sdOihk2uxE/Unilever---Snack-snap?content-scaling=fixed&kind=proto&node-id=48-4510&page-id=1%3A2&scaling=scale-down&starting-point-node-id=48%3A4510&embed-host=cawi&client-id=Ptr4uxHDgFfqLl6vjtttEPPrototype is shared public. Demo https://cawi.nms.cz/3e1f3e8?Expected Embed API works like this without be logged into figma.When I am logged in:But in anonymous window (without logged into figma)

I’m an admin for our Organisation and am trying to assess the usage of AI credits across the team before the charges begin (as per the recommendation from Figma). However, I can only view team member’s usage from this current payment month which has only just begun. How do I see the usage from last month so I can see what the usage was at the end of the month to be able to understand how many credits we typically use?

URGENT , j’ai besoin d’aide pour connecter mon application WEB à ma boite email professionnel ainsi qu’un accès à la plateforme via le site web ?Mon objectif est d’ajuster la bonne coordination entre mon application web à ma boite email professionnel  sous figma avant lancement des tests , publication de mon application au niveau USA . Merci pour votre retour

I’ve been involved in a few projects recently where choosing the right annotation partner made a huge difference in downstream model performance. It’s surprising how two vendors that look similar on the surface can behave very differently once you get into real workloads.Some of the differences I keep noticing include: consistency of labels across reviewers how edge cases are handled clarity and completeness of guidelines how errors are tracked and corrected transparency in the whole workflow One write-up I found explained a lot of these nuances pretty clearly:https://aipersonic.com/blog/choosing-the-right-annotation-partner/Sharing that just for context — not as a recommendation.For those of you who’ve actually hired annotation teams before:What were the most important factors for you in choosing a partner?Did accuracy matter more than communication?Did workflow transparency make a difference?Did you test with a small pilot first?Would love to hear your real experiences and ti

For smaller datasets, annotation can feel straightforward. But once you push past a few tens of thousands of samples, a bottleneck seems to form almost out of nowhere — quality drops, inconsistency spikes, and QA starts lagging.A few common pain points I keep running into: annotation drift between reviewers edge cases that weren’t covered in initial guidelines QC that worked early but collapses under volume scaling human reviews without losing quality I was thinking about where these bottlenecks come from and what steps actually help break through them. This write-up I read framed a lot of the pain points and mitigation steps in a way that made sense to me:https://aipersonic.com/blog/breaking-the-annotation-bottleneck/Not sharing the link to promote anything — just to give some context for the discussion.So for folks who’ve dealt with this at scale:What actually helped you lessen the bottleneck?Was it:• better reviewer training?• multi-stage QA?• automation + human hybrid?• clear

Working with medical data really changed how I think about annotation. Compared to general image or text labeling, the margin for error feels much smaller, and the consequences of inconsistency show up much faster.Some things that stand out to me: tiny annotation differences can change clinical meaning edge cases are common, not rare reviewer agreement is harder to maintain QA needs to be much stricter than usual A lot of problems don’t appear until later, when models are validated or retrained, which makes them harder to trace back to annotation decisions.Looking at structured medical annotation workflows helped me understand where these issues usually come from and how teams try to control them:https://aipersonic.com/medical-annotation/Sharing this for context, not as a recommendation.For those who’ve worked on medical AI projects: how do you manage reviewer consistency? do you rely fully on domain experts, or a mix of trained annotators + review? what QA steps actually mad

When i try to log in the account, it shows that “You've made too many recent attempts. Please try again later.”. What can i do to successfully log in? How long do i need to wait? Any other solutions to fix the problem?

Once I do update on figma i can push to github then if I work locally as developer and send push to github then from figma ai how can i pull those updates codes ?

I have design system and buttons status count too big and need to use variables but need some help

I’m trying to get Cline to ingest Figma files for a variety of use cases. One is to have it run a checklist of design library guidelines on a set of screens that are nearly ready to be passed to Dev. Or, have Cline be a chatbot assistant that understands a very large service blueprint. However, whenever I give it the Figma files, it always locks up as there is way too much information in the files for the model to take on. I’ve tried to parse or chunk this out but nothing seems to work. Can anyone help me understand what I can do differently? I’m using the MCP server, and saw from a YouTube video (22 mins mark) that the MCP should have a way to auto chunk this but I cannot get this to work. TIA

I’m currently setting libraries for multiples teams, all them have the same configuration to be shared only for the team, but there is only one that keeps showing in all teams libraries even with all the “shared” options turned off. I already unpublished and published for team only, but the issue persists. Am I doing something wrog? 

how to remove figma watermark

I cant get the @ on my figma website to connect to godaddy

Hello,I have been unable to use Figma for 5 days already.I constantly get the Connection error: -106, and files do not load.What I have already tried: My internet connection is stable and fast Tried different networks Reinstalled Figma Tried both browser and desktop app Restarted my computer multiple times Nothing has helped so far.Please let me know what is causing this issue and how I can fix it, as I am currently unable to work.Thank you in advance.

Hello,I have accidentally linked Figma Education to the wrong account - I even went ahead and approved it from my student email address via SheerID 🤦🏻I have sent an email 2 weeks ago asking for support but I am not getting any answers. Can you kindly help me?

I have installed figma on PC and created a account and after trying so many times I'm not able to login via email and not receiving authentication code even after so many times 

Hi everyone,I’m trying to publish a Figma plugin that provides a list of UI and Design System items.The idea is simple: users see a list of components inside the plugin UI, and when they click on an item, that component is inserted into their canvas as an Instance.The important part is that these items are components defined in a separate file or library, and the plugin only inserts instances of them into the user’s canvas. When the user clicks “Go to main component”, they should not have access to the original library file.I don’t really care whether the component keys are hard-coded in the plugin or auto-updated.The real problem is that I can’t even list or access those components inside the plugin.From what I understand after reading the Figma API documentation: Component keys from a published library are only accessible to members of the same team. Figma does not grant public plugins permission to read or enumerate components from private or team-restricted libraries. There is

I’m a freelance developer working with a client. The client has shared Figma files with me. I have a “professional” subscription with the intent to use dev mode to help me build the project.When I hit the dev mode button, I’m told I have to request access.Previously when I tried this, before I had my own professional account, I was told they couldn’t grant me access as there wasn’t a seat for me on their plan.If I now request access, will it use my seat / professional account to give me dev mode access to the project files, or will it insist on the client granting me a dev mode seat on their account? (if the latter, how do I go about requesting a refund for my professional account, as that was the whole point of me signing up this morning).