Loading ToolBench
Japan Design — UI + UX correctness for AI coding agents. Enforces Japanese front-end (typography, colour, layout) and UX (forms, keigo, trust) conventions on real CSS/markup, with JLReq/JIS citations.
Japan Design MCP has 12 tools with strong domain-specific descriptions (avg 180 chars, well above baseline 72 chars for params). All tools have clear, actionable descriptions explaining WHAT they do and WHEN to use them. Input schemas are present and use proper Zod validation with enums and type constraints. However, output schemas are NOT documented, tools return JSON but LLMs cannot see the response structure in advance. Tool naming is verb-forward (review_, generate_, validate_, suggest_, score_, transform_, check_, audit_, design_, get_) and follows conventions. One critical issue: tool #1 (review_jp_ui) and #12 (review_jp_ui) are DUPLICATES with identical names, descriptions, and schemas, this will confuse LLM tool selection. Parameters are well-constrained with enums (context, format, language, ui_element, site_type) and include helpful descriptions. Error handling is not visible in the provided code, tool implementations are imported but not shown, so recovery guidance cannot be verified.
Broad 7-category UX audit: layout (grid, spacing, density, responsive), trust (company info, contact, legal), navigation (structure, findability, breadcrumbs), content (copy tone, keigo, density), mobile (touch targets, viewport, performance), typography (font, size, line-height), and cultural (seasonal, imagery, accessibility). Returns score (0-100), category breakdown, and priority fixes.
Deep typography-only audit: font stack (serif/sans-serif, Japanese font support, fallback chain), font size (base size, scaling, readability for 65+ demographic), line-height (leading, density, JLReq compliance), letter-spacing (tracking, kinsoku rules), and text rendering (font-smoothing, kerning). Returns violations with JLReq/JIS citations and fixes. Subset of review_jp_ui focused on type.
Generate a Japan visual brief from brand/audience/industry: recommended font stack (serif/sans-serif, Japanese fonts, fallbacks), colour palette (seasonal, cultural, accessibility), layout pattern (grid, spacing, density), typography scale (sizes, weights, line-heights), and component style guide (buttons, forms, cards). Returns a structured design system brief with JLReq/JIS citations.
Generate culturally correct Japanese form markup with proper field order (姓→名), furigana, 3-field phone, 〒 postal address, 年月日 dates, and context-appropriate keigo. Use this when building any form for a Japanese audience.
DUPLICATE TOOL: review_jp_ui appears twice (tools #1 and #12) with identical name, description, and schema. LLM will not know which to call and may fail to select either.
Output schemas NOT documented. Tools return JSON (e.g., review_jp_ui returns verdict/violations/fixes, generate_jp_form returns markup/notes) but LLMs cannot see the response structure in advance. This forces LLMs to guess field names and types for downstream reasoning.
Error handling and recovery guidance not visible in source. Tool implementations are imported but not shown; cannot verify whether errors include actionable guidance (e.g., 'Invalid context: must be one of...') or just fail silently.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 61 | 2026-07-28+ | v2 |
Generate realistic Japanese test data for prototypes and development. Returns names (kanji + katakana furigana + romaji), addresses (real postal codes and prefectures), phone numbers (correct 3-field format), company names, and dates (both Gregorian and Japanese era format like 平成4年3月15日).
Return seasonal colours, events, and launch-blackout dates for Japan. Includes: current season (春/夏/秋/冬), seasonal colour palette (traditional Japanese colours like sakura-iro, matcha-iro), major holidays and events (New Year, Golden Week, Obon, year-end), and launch-blackout dates (avoid launches during major holidays or peak shopping periods). Useful for timing campaigns and choosing seasonal design directions.
Review CSS/markup against the Japanese UI standard (typography, colour, font performance) and return a pass/needs_work/fail verdict with violations, JLReq/JIS citations, and fixes. The front-end correctness enforcer — run on any UI (including Western-built) before shipping to Japan; call transform_for_japan to auto-apply form/copy fixes.
Review CSS/markup against the Japanese UI standard (typography, colour, font performance) and return a pass/needs_work/fail verdict with violations, JLReq/JIS citations, and fixes. The front-end correctness enforcer — run on any UI (including Western-built) before shipping to Japan; call transform_for_japan to auto-apply form/copy fixes.
Score any page or component for Japan-readiness on a 0-100 scale. Analyzes 5 categories: forms (name order, furigana, phone, postal), copy (keigo, Japanese text, placeholders), trust (特定商取引法, phone number, company info), typography (font, size, line-height), and cultural (seasonal awareness, imagery). Returns breakdown with issues and quick wins per category.
Suggest appropriately polite Japanese UI text based on context. Maps business context to keigo level (casual → very_formal) and returns the right Japanese translation for buttons, error messages, empty states, confirmations, and more. Includes alternatives for different formality levels.
Auto-apply Japanese UX fixes to Western markup: reorder name fields (姓→名), add furigana fields, convert phone to 3-field format, add postal code field, convert dates to 年月日, upgrade button text to context-appropriate keigo, add required field indicators, and add placeholder examples. Returns fixed markup and a list of changes made.
Audit an existing form against Japanese UX conventions. Checks for: name field order (姓/名), furigana, 3-field phone, postal code auto-fill, keigo-appropriate button text, full-width character handling. Returns score (0-100), issues with fixes, and what passed.
No pagination or result limits documented. Tools like generate_jp_placeholder accept count up to 50, but audit_japan_ux and score_japan_readiness do not specify max response size. Large markup inputs could produce unbounded output.
Tool composition: no clear guidance on which tool to call first. Instructions mention 'Start here for is this UI correct for Japan?' (review_jp_ui) but do not explain when to use audit_japan_ux vs review_jp_ui vs check_jp_typography. LLM may waste calls on overlapping tools.