figma-check
Skill1 starsChanged 16 days agoMITView on GitHub
Downloads node and render from Figma and compares them with the app's real screen, measurements included. Recognises an expired token in a single call, and when the design looks wrong it stops and asks instead of deciding on its own.
In this repository5 other files it gives its agents.
- deliverSkill
- gui-testSkill
- squadraSkill
- start-workSkill
- verify-for-realSkill
The file
We split this skill into the rules it gives an agent
7 rules. About 0 are common, 3 are distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 24
any: call, /v1/me included: try that one first and you'll know in one shot whether it's the token or the permissions.
- Very specific: line 30
If you only have the screen's name, walk down level by level instead of guessing: ?depth=1 gives the pages, ?depth=2 the sections, then the …
- Very specific: line 35
Put the render next to the simulator capture (gui-test) and look at text, sizes and order, not the general impression.
- Very specific: line 39
Watch out for truncated text: If a title ends with an ellipsis, check textTruncation, maxLines and the box width: if they aren't there, the …
- Distinctive: line 45
Stop and ask before implementing: Don't invent copy or measurements, and don't silently ship the version that looks right to you.
- Distinctive: line 48
Don't block everything else: implement what doesn't depend on that answer, hold back the uncertain piece, and ask at the right moment.
- Distinctive: line 51
If the user decides to deviate from the node, write it down where someone will find it: a code comment with the node id and the …
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Comparing with Figma
Access
The Figma personal access token lives in an environment variable (FIGMA_TOKEN),
never in the repo. The file key lives in the project docs. A project can have
more than one file: asking the wrong file for a node returns null without
saying why, and it's the first place time gets lost.
curl -sH "X-Figma-Token: $FIGMA_TOKEN" "https://api.figma.com/v1/files/$KEY/nodes?ids=123:456"
curl -sH "X-Figma-Token: $FIGMA_TOKEN" "https://api.figma.com/v1/images/$KEY?ids=123:456&format=png&scale=2"Ids use - in URLs and : in the API. The render comes back as a short-lived
URL: download it right away.
The token expires. The signature is {"status":403,"err":"Token expired"} on
any call, /v1/me included: try that one first and you'll know in one shot
whether it's the token or the permissions. When it has expired, don't look for
workarounds: ask the user for a new token.
Finding the node
If you only have the screen's name, walk down level by level instead of guessing:
?depth=1 gives the pages, ?depth=2 the sections, then the frames inside the section.
Comparing
Put the render next to the simulator capture (gui-test) and look at text,
sizes and order, not the general impression. For text, read characters from
the node: that's the source, not what you read in the render.
Watch out for truncated text. If a title ends with an ellipsis, check
textTruncation, maxLines and the box width: if they aren't there, the
designer typed the dots and it isn't an elision. It happens.
When the design looks wrong
Stop and ask before implementing. Don't invent copy or measurements, and don't silently ship the version that looks right to you.
Don't block everything else: implement what doesn't depend on that answer, hold back the uncertain piece, and ask at the right moment.
If the user decides to deviate from the node, write it down where someone will find it: a code comment with the node id and the measurement, the docs, and a test that fails if it goes back. Otherwise the next person who rereads Figma "restores" it in good faith.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

