Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add alvinindra/figma-mcp-rust --skill read-design-strategygit clone --depth 1 https://github.com/alvinindra/figma-mcp-rustWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/alvinindra/figma-mcp-rust/read-design-strategy)<a href="https://agentmods.dev/skills/alvinindra/figma-mcp-rust/read-design-strategy"><img src="https://agentmods.dev/badge/skills/alvinindra/figma-mcp-rust/read-design-strategy/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/alvinindra/figma-mcp-rust/read-design-strategy"><img src="https://agentmods.dev/badge/skills/alvinindra/figma-mcp-rust/read-design-strategy.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00057 | $0.00572 |
| Opus 5 | $0.00028 | $0.00286 |
| Sonnet 5 | $0.00011 | $0.00114 |
| Haiku 4.5 | $0.00006 | $0.00057 |
Grade A, and why
read-design-strategy scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 10d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
To effectively read a Figma design with figma-mcp-rust:
- Start with get_metadata to understand file name, pages, and current page
- Use get_pages to list all pages without loading their full trees
- Use get_design_context (depth=2, detail=compact) for a token-efficient summary of the current selection or page
- detail=minimal: id/name/type/bounds only (~5% tokens)
- detail=compact: + fills/strokes/opacity (~30% tokens)
- detail=full: everything, default (100% tokens)
- dedupe_components=true: INSTANCE nodes are collapsed to compact stubs (mainComponentId + componentProperties overrides); unique component structures are collected once in a top-level componentDefs map. Use this whenever the screen contains repeated component instances (e.g. card lists, table rows, nav items). Typical savings: 5-10x fewer tokens vs full serialization of repeated instances.
- For screens with many repeated components, the recommended reading flow is: a. get_design_context(depth=2, detail=minimal, dedupe_components=true) to see the instance layout and component IDs b. Inspect componentDefs in the response: one definition per unique component, not one per instance c. Read componentProperties on each instance stub for variant selections, text overrides, and boolean toggles d. Drill into specific instances with get_node only when an instance has unique overrides you need to inspect
- Use search_nodes to find nodes by name or type without dumping the entire tree
- Drill into specific nodes with get_node or get_nodes_info (prefer batch over single calls)
- For text-heavy components, use scan_text_nodes to collect all copy at once
- Use scan_nodes_by_types to find all FRAME/COMPONENT/INSTANCE nodes in a subtree
- Call get_styles and get_variable_defs once per session to understand the design system
- Call get_fonts to understand typography usage across the page at a glance
- Use get_viewport to see what the user is currently looking at in the canvas
- Use get_reactions to inspect prototype interactions on a node
- Call get_screenshot last and only when visual confirmation is needed, because it is expensive
- Node IDs use colon format (4029:12345), never hyphens
- get_local_components returns componentSets and variantProperties for variant-aware inspection
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 10d ago First seen · 34 lines · 57 tokens per session scan A 075085873c73
read-design-strategy is a skill published in the GitHub repository alvinindra/figma-mcp-rust (38 stars, last pushed 6d ago), licensed MIT. It adds 57 tokens to every session and 572 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
clean-interface-design
Design clean, calm, uncluttered interfaces that don't read as templated AI output. Use when building or restyling any UI (web, iOS, Android, macOS) and it needs visual hierarchy, restraint, spacing discipline, one clear primary action, and a cohesive look. Covers accent color, whitespace, corner radius, type scale…
landing-page-conversion
Build or improve landing pages and marketing sites that convert, not just look good. Use when writing a hero, structuring a page, choosing a CTA, adding social proof, or diagnosing why a page isn't converting. Covers positioning, copy-first workflow, the above-the-fold contract, page narrative, CTAs, trust, and…
ship-quality-gate
Run the deterministic auditors over a real project before it ships — design drift, security headers, generated-default tells, SEO/GEO signals, delivery signals, Apple and Android project configuration — and read the results correctly, disclosure list included. Use when someone asks "audit this", "is this ready to…
apple-platform-design
Design iOS, iPadOS, and macOS apps that feel native, in the Liquid Glass era. Use when building or reviewing Apple-platform UI — navigation, controls, sheets, materials, typography, haptics — or deciding what must change when porting from web/Android. Covers Apple's HIG, the Liquid Glass design language (iOS 26 /…
design-system-audit
Find out whether a codebase actually has a design system or is re-deciding the basics on every screen — count the distinct colors, sizes, radii, shadows and spacings, collapse the near-duplicates, and migrate onto tokens. Use when inheriting a codebase, before a redesign, when a UI "feels inconsistent but I can't say…
motion-and-animation
Add motion and animation that feels great, not like slop. Use when animating any UI — transitions, hovers, presses, sheets, drawers, modals, page transitions, gestures — or when an animation feels "off". Covers the easing rules agents get wrong (ease-out for enter, ease-in for exit), springs, interruptibility…