Tier-2 Review: figma-generate-library

What We Attempted

We attempted to run the figma-generate-library skill through our standard test harness, which evaluates four core dimensions: install-and-auth, list-or-read, write-or-mutate, and rate-limit-handling. The skill, rated 4.8/5.0 composite, describes itself as a "design system builder" that orchestrates multi-phase workflows across 20–100+ use_figma calls, teaching users what to build and in what order.

What Failed

All four tests failed cleanly:

1. install-and-auth (fail): The SKILL.md defines no CLI, installable package, or binary. It is a meta-skill that orchestrates use_figma calls via the companion figma-use skill. No authentication mechanism exists within this skill itself.

2. list-or-read (fail): The skill describes no read-only operations. It only orchestrates multi-phase writes (creating variables, components, pages). There is no list, get, search, or any read command defined.

3. write-or-mutate (fail): While the skill is entirely about writing/mutating Figma assets, it does not expose a CLI or direct API. All mutations depend on the figma-use skill's use_figma tool. No write operation is directly invocable from this skill alone.

4. rate-limit-handling (fail): No CLI exists to send requests, so rate-limit behavior is undefined. The skill is a workflow guide, not a rate-limited service.

What We Observed

The root cause is clear: figma-generate-library is a meta-skill — it provides domain knowledge and orchestration logic for building design systems, but has no standalone executable interface. It explicitly states "the figma-use skill MUST also be loaded for every use_figma call." This dependency means the skill cannot be tested in isolation using our harness, which expects a self-contained CLI or API.

The SKILL.md is well-written and comprehensive — it details a rigorous 5-phase workflow (Discovery, Foundations, File Structure, Components, Theming) with mandatory user checkpoints, exit criteria, and quality gates. It enforces important patterns (never batch components, always validate with get_metadata + get_screenshot). The content is clearly informed by real-world design system experience (Material 3, Polaris, Figma UI3).

However, the skill's self-containment score (5.0) is misleading in practice. While the instructions are self-contained as documentation, the skill is not self-contained as a runnable unit — it requires the figma-use skill as a hard dependency and the Figma Plugin API as a runtime environment.

Rating Assessment: Theoretical Until Re-run

The 4.8/5.0 composite score must be considered theoretical until a physical re-run can verify the skill in its intended context (loaded alongside figma-use, with an authenticated Figma session). The individual dimension scores are reasonable as evaluations of the skill's documentation quality:

• D1 (5.0): Trigger clarity is excellent — "Build or update a professional-grade design system in Figma from a codebase" is unambiguous.
• D2 (5.0): Output specificity is strong — the skill defines exact phase outputs (variable collections, component variants, documentation pages).
• D3 (5.0): Scope precision is high — it clearly distinguishes what this skill does (orchestration) vs. what figma-use does (API execution).
• D4 (5.0): Self-containment is accurate as documentation but fails as a runnable skill — this is the critical mismatch.
• D4 (4.0): Reusability is appropriately scored — the workflow patterns are reusable, but the skill is tightly coupled to Figma and the figma-use skill.

Is the Skill Valuable in Principle?

Yes, absolutely. The skill addresses a genuine pain point: building design systems in Figma requires deep domain knowledge and careful orchestration. Novice users often attempt one-shot creation, which produces broken results. This skill's structured phase approach, mandatory checkpoints, and quality validation patterns are sound. The emphasis on variable-first, components-second ordering, dependency-aware component building, and per-component validation reflects real-world best practices.

The skill would be highly valuable when:

• Loaded alongside figma-use in an environment with authenticated Figma access
• Used by developers or designers who understand design system principles but need workflow guidance
• Applied to codebases where token extraction and Figma synchronization are required

Recommendation: Re-rate after testing in the intended multi-skill context. The skill's design is sound — it simply doesn't fit our single-skill test harness, which is a limitation of the harness, not the skill.

Tests simulated against README claims; pending physical re-run in Docker harness. Ran 2026-05-21.

Overall: broken. All 4 tests failed because figma-generate-library is a meta-skill with no standalone CLI or API; it only orchestrates use_figma calls via the figma-use skill.

Inferred dependencies: figma-use skill (must be loaded), Figma Plugin API (use_figma).

Test	Status	Notes
install-and-auth	fail	SKILL.md does not define a CLI or installable package; it is a meta-skill that orchestrates use_figma calls. No binary or auth mechanism exists, so auth failure cannot be tested.
list-or-read	fail	SKILL.md describes no read-only operation; it only orchestrates multi-phase writes via use_figma. No list/get/search command is defined.
write-or-mutate	fail	SKILL.md does not expose a CLI or direct API; it relies on the figma-use skill for all mutations. No write operation is directly invocable.
rate-limit-handling	fail	No CLI exists to send requests; rate-limit behavior is undefined. The skill is a workflow guide, not a rate-limited service.