figma-readiness
SkillDocs & knowledgeScore a Figma frame's structure before building from it, default layer names, hidden leftovers, width in the frame name, components, text as text, tokens, behaviour notes, and write the message to send the designer if it is not ready. Use before the first get_design_context call of any build, and whenever asked whether a frame is buildable or why a build keeps needing re-prompts.
Instructions available. Your AI can read the instructions. Execution depends on the setup they require.
Account requirements not reviewed. Check the skill instructions before use; ahel provides instructions and does not run this skill.
Add ahel to your AI once: Claude, ChatGPT, Cursor, Claude Code or Codex. Then ask it to use this.
Then ask your AI: use the figma-readiness skill
What this skill tells your AI
The instructions your AI receives, as published by ecomexperts-io/base in .claude/skills/figma-readiness/SKILL.md and read by ahel’s review.
Check the frame before spending a day on it.
Why this exists. The designer practices document lists sixteen things and
calls itself aspirational. Every re-prompting cycle it describes traces back to
a few structural facts a script can see: layers named Frame 1261153641,
hidden old versions, no width in the frame name, no components, text flattened
to vectors. Until v2 nothing checked a frame before a build started, so the
questions went back to the designer after the build, one at a time. This sends
them back before, all at once, with the specific layers named — which is how
"this is what the workflow requires" gets said to a designer: as a check with
numbers, every time.
Invocation: /figma-readiness <fileKey> <nodeId>[@<width>] [<nodeId>@<width>]
— runs as Step 1½ of /build-page-from-figma; also on its own.
Steps — per frame
-
Metadata. Call the Figma
get_metadatatool with the file key and node id. Save the XML it returns to.claude/verify/readiness/<nodeId>.xml(write the response body as-is). -
Score.
python3 .claude/scripts/figma-readiness-score.py .claude/verify/readiness/<nodeId>.xml --width <width>It prints a verdict per check —
pass,warn,ask,send back— the largest default-named layers to rename first, and any hidden layers. -
The two things metadata cannot show.
- Auto layout. Call
get_design_contexton the frame (invoke the Figma design-to-code skill first, as the build skill requires). If the returned layout is dominated by absolute positioning rather than flex/grid, that is asend backon practice 6, whatever the score said. - Tokens. Call
get_variable_defs. Empty means colours, spacing and type will be reverse-engineered from raw values — awarnto state in the report and to the designer. - Behaviour. For any interactive component (menu, drawer, accordion,
filter, carousel) look for text layers or comments beside the frame that
describe states. None →
askbefore building the behaviour.
- Auto layout. Call
-
Second breakpoint. If only one node id was given, ask for the other. Two frames with the width in each name is practice 3; a single frame means the other breakpoint will be invented.
Verdict
- pass / warn — build. Put the warnings in the build report so the designer sees them once.
- ask — build what is unambiguous; ask the specific question before the ambiguous part.
- send back — do not build yet. Write the designer a short message: the verdict lines, the layers to rename (ids and names), the hidden layers to remove, the missing width or breakpoint, in plain language and under twenty lines. Hand it to the developer to forward. Then stop — a developer can override with "build anyway", and the report records that they did.
What to record
Copy the scorer's output into the build report under "Frame readiness". A
frame that scored send back and was built anyway is a fact the reviewer
should know; a frame that scored pass and still needed three re-prompts is
an incident for this skill (/record-incident with should_have_caught
pointing here).
Signals
- GitHub stars
- 28
- Forks
- 16
- Last commit
- Oct 2026
ahel review
S4info
community integration, published by ecomexperts-io, not figma
Automated review, not a security audit. Ruleset v1+k2.
Advanced
- Item type
- skill
- Key
figma-readiness- Source
- github.com/ecomexperts-io/base
github.com/ecomexperts-io/base
Related picks
Skill · heygen-com
The pick for Figmahandoff
Skill · mattpocock
More in Docs & knowledgecanvas-design
Skill · anthropics
More in Docs & knowledgedoc-coauthoring
Skill · anthropics
More in Docs & knowledgewriting-for-agents
Skill · mattpocock
More in Docs & knowledgespec-driven-development
Skill · addyosmani
More in Docs & knowledge