notque/vexjoy-agent / skills/frontend/jev-design
jev-design
VerifiedDesign React interfaces with the full shadcn/ui catalog. The agent writes a brief, Jev picks components and a style recipe from bounded catalogs, then the agent implements, renders, and reviews against concrete checks.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 7.4 KB |
| references/brief-writing.md | markdown | 3.3 KB |
| references/catalog-v1.json | config | 2.3 KB |
| references/catalog-v2.json | config | 48.9 KB |
| references/decision-card.md | markdown | 2.5 KB |
| references/jev-images.md | markdown | 3.9 KB |
| references/visual-review.md | markdown | 3.3 KB |
| scripts/plan.py | script | 28.2 KB |
| tests/test_plan.py | script | 17.3 KB |
references/decision-card.md
View on GitHub# Decision card
- Unit: one bounded screen brief with up to eight named requirements.
- Baseline: the tightest-fit compatible component per requirement (fewest extra capabilities, then non-legacy, then catalog order) and the recipe with the most requested traits.
- Judgments: one Choice per requirement, one Choice for the complete style recipe, and an independent absolute-fitness Noul for every candidate.
- Policy: accept only in-catalog choices whose fitness is at least `0.62` (provisional). Otherwise abstain and return the baseline and candidates.
- Failure: missing, partial, malformed, non-finite, or out-of-range answers abstain. So do an unavailable transport, any failed request, and storage failures. Partial answers from a split run are never used.
- Bounds: 1,500-character brief (down from 12,000, because the brief rides in every request), eight requirements, eight capabilities per requirement, six candidates per requirement, ten style candidates, and six style traits.
- Requests: each requirement's candidates are packed into requests of 3,500 estimated tokens or fewer, plus one style request. Nothing over 4,500 tokens is sent. A test covers the worst case: 8 requirements on the three most expensive capabilities, maximum field lengths, 4 requests, about 10.5k tokens per run, 3,266 tokens in the largest request.
- Retries: the transport retries 429, 503, and 529 with jittered backoff, 4 attempts at most. The planner makes one round of requests. The budget check warns only when it assumes every request retries 4 times within one latency window.
- Versions: served model `typesafe-ai/jev` through Vercel AI Gateway (unversioned; the served model is recorded per attempt). The v1 pin `jev-1.13.0` applied to the direct API only. Rubric `jev-design-rubric-v2`, policy `jev-design-policy-v2`, catalog `shadcn-style-catalog-v2`. The catalog has 65 components (all 63 `registry:ui` items from `https://ui.shadcn.com/r/index.json` plus the `data-table` and `date-picker` composites), 59 capabilities, and 10 recipes, fetched 2026-09-22.
- Threshold note: `0.62` was set on v1 wording. The v2 questions reference state paths and drop Noul criteria, so recalibrate on labeled briefs before promotion.
- Falsifier: stay advisory until group-disjoint human labels show at least 80% accepted component plans, with no deterministic incompatibility accepted.
- Safety: the planner cannot install components, edit product files, verify accessibility, inspect screenshots, or establish visual quality. Recipe contrast is computed from its OKLCH values, not measured on a render.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| brainstormingobra/superpowersYou MUST use this before any creative work - creating features, building components, adding functionality, modifying behavior, or planning anything new, in software or out of it (a talk, a business, a renovation). | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| brand-guidelinesanthropics/skillsApplies Anthropic's official brand colors and typography to any sort of artifact that may benefit from having Anthropic's look-and-feel. Use it when brand colors or style guidelines, visual formatting, or company design standards apply. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| frontend-designanthropics/skillsGuidance for distinctive, intentional visual design when building new UI or reshaping an existing one. Helps with aesthetic direction, typography, and making choices that don't read as templated defaults. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| mcp-builderanthropics/skillsGuide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK). | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |