Skip to main content
Glama
ReftrixMCP
by TKMD

ReftrixMCP

Web design knowledge base platform -- layout analysis, motion detection, and quality evaluation via MCP tools.

For frontend engineers, designers, and AI-agent builders who want to analyze real websites and retrieve reusable UI patterns via Claude or any MCP client.

License: AGPL-3.0-only TypeScript PostgreSQL MCP pnpm

ReftrixMCPは、Webデザインパターンをベクトル検索(pgvector HNSW)と RAGで検索可能なナレッジベースに集約し、MCPツール経由でClaude等の AIエージェントと統合するプラットフォームです。

主要機能: レイアウト分析 / モーション検出 / 品質評価 / セマンティック検索 / 横断検索 / 画像類似検索 / レスポンシブ解析 / パーツ分析 / レート制限 / 検索キャッシュ / BullMQ UI / SBOM

28のMCPツールを提供: Layout(4) / Motion(2) / Quality(1) / Page(3) / Narrative(1) / Background(1) / Responsive(2) / Part(3) / System(1) / Search(1) / Design(3) / Data(2) / Audit(1) / Accessibility(1) / Performance(1)

詳細な日本語ドキュメント: docs/README.ja.md

What it does

  • Layout analysis -- auto-detect sections (hero, feature, CTA, etc.), extract grid/typography, and generate React/Vue/HTML code

  • Motion detection -- discover CSS/JS animations with frame capture (15 px/frame video mode), CLS detection via Pixelmatch

  • Quality evaluation -- score designs on three axes (originality, craftsmanship, contextuality) with anti-AI-cliche detection

  • Semantic search -- find layout, motion, narrative, background, and responsive patterns via pgvector HNSW hybrid search

  • Part-level analysis -- extract 16 UI part types (button, icon, heading, etc.) with DINOv2 visual embeddings for visual similarity search

  • Vision integration -- Ollama llama3.2-vision for richer layout, motion, and narrative understanding

  • Section post-processing -- auto merge/split sections by type, heading, and height (Rule 1-4) for optimal structure

  • Multi-tile capture -- split large sections (>viewport height) into tiles for complete DINOv2 visual coverage

  • Blank image detection -- detect lazy-loading unrendered sections and re-capture via Playwright for full coverage

  • Code generation -- convert analyzed sections to React, Vue, or plain HTML with matched motion patterns

  • Unified search -- cross-service search across layout, part, motion, background, and narrative patterns in a single query

  • Image similarity search -- find visually similar designs via DINOv2 embeddings from Base64/URL input (RRF 3-source)

  • Rate limiting -- Token Bucket + Redis Lua (CWE-770 DoS prevention), 3-tier (analysis 10 RPM / search 120 RPM / default 60 RPM)

  • Search cache -- LRU in-memory cache (lru-cache v11) with TTL-based natural expiry (5 min)

  • BullMQ UI -- Bull Board dashboard for monitoring async page.analyze jobs (port 21080)

  • SBOM -- CycloneDX 1.6 auto-generation for EU CRA vulnerability reporting compliance

Related MCP server: Firecrawl MCP Server

Why ReftrixMCP

Layout-aware

Sections, grids, and typography extracted as structured data -- not just screenshots

Motion-aware

CSS static analysis + frame-by-frame video capture for real animation behavior

Quality-aware

Three-axis scoring with actionable improvement suggestions

Searchable

768-dim multilingual embeddings (e5-base) with HNSW index and hybrid RRF ranking

Preference-aware

Search tools accept an optional profile_id and rerank with an existing profile from an older version; profile creation was removed on 2026-11-04

Part-aware

16 UI part types extracted with DINOv2 visual embeddings for cross-site component comparison

MCP-native

28 tools purpose-built for Claude Desktop and MCP Client CLI

Quickstart

Run page.analyze on any URL in under 5 minutes.

Prerequisites

Node.js ^22.23.4 || ^24.22.1 || ^26.12.1, pnpm 10+, Docker & Docker Compose, Ollama

Setup

git clone https://github.com/TKMD/ReftrixMCP.git && cd ReftrixMCP
pnpm install                                     # CUDA skip is default; see GPU note below
cp .env.example .env.local                       # edit DATABASE_URL / REDIS_URL as needed
cp .env.local packages/database/.env             # Prisma CLI requires this copy
pnpm docker:up                                   # PostgreSQL 18 + pgvector + Redis
pnpm db:migrate && pnpm db:seed
pnpm build
pnpm exec playwright install chromium            # browser for page crawling
pnpm --filter @reftrixmcp/ml download:dinov2        # DINOv2 visual embedding model (~330 MB)
pnpm --filter @reftrixmcp/ml repair:e5-cache --check # (optional) verify multilingual-e5-base ONNX cache (~1.1 GB) integrity
curl -fsSL https://ollama.com/install.sh | sh    # install Ollama
ollama pull llama3.2-vision                      # vision model (~7.9 GB)
ollama serve                                     # keep running in a separate terminal

Note: If you change .env.local, also update packages/database/.env. page.analyze workers are auto-forked by WorkerSupervisor when the MCP server starts (v0.4.0 PR7d-2+). Manual start via pnpm --filter @reftrixmcp/mcp-server worker:start:page is developer-only and requires REFTRIX_ALLOW_MANUAL_WORKER=true to bypass the Redis-based dual-run guard if the MCP server is also running. See Getting Started for GPU configuration and details.

GPU / CUDA: CUDA binary download is skipped by default (CPU fallback). For GPU acceleration setup, see Troubleshooting: CUDA Detection.

pnpm "Ignored build scripts": On pnpm install, pnpm 10 prints Ignored build scripts: @prisma/client, esbuild, sharp, ... and suggests Run "pnpm approve-builds". This is expected, not an error — pnpm 10 blocks dependency build/lifecycle scripts by default as a supply-chain safeguard. Setup still completes because sharp and esbuild ship prebuilt binaries, and the Prisma client is generated by the @reftrixmcp/database workspace package's own postinstall (prisma generate) during pnpm install — a workspace lifecycle script, which pnpm does not gate behind the dependency build-script allowlist. Only run pnpm approve-builds if you have a specific reason to let one of these blocked dependencies run its own native build step (onnxruntime-node is already allow-listed via pnpm.onlyBuiltDependencies, so it never appears in the ignored list; and CUDA acceleration is a separate opt-in — see the GPU / CUDA note above — not pnpm approve-builds).

Using the npm package

If you install @reftrixmcp/mcp-server from npm instead of cloning, the Playwright browser is not included: get it with npx playwright@1.64.1 install chromium (any directory). See "Using the npm Package" in the MCP Server README for the browser notice (third-party software, not covered by this project's license, not redistributed, not in the SBOM) and the --ignore-scripts step (npm rebuild @prisma/client).

Connect to Claude

Add to your MCP config:

  • Claude Desktop: ~/Library/Application Support/Claude/claude_desktop_config.json (macOS)

  • MCP Client CLI: .mcp.json in the project root or ~/.claude/.mcp.json

{
  "mcpServers": {
    "reftrix": {
      "command": "node",
      "args": ["/absolute/path/to/ReftrixMCP/apps/mcp-server/dist/index.js"],
      "env": {
        "NODE_ENV": "development",
        "DATABASE_URL": "postgresql://reftrix:change_me@localhost:26432/reftrix?schema=public",
        "REDIS_URL": "redis://localhost:27379",
        "OLLAMA_BASE_URL": "http://localhost:11434",
        "OLLAMA_HOST": "http://localhost:11434",
        "ENABLE_SECTION_SCREENSHOT_FALLBACK": "true"
      }
    }
  }
}

Replace change_me with a secure password. Port 26432 = standard 5432 + 21000 offset.

OLLAMA_BASE_URL is used by the MCP server process; OLLAMA_HOST is used by the worker process. Both must match if Ollama runs on a non-default port.

ENABLE_SECTION_SCREENSHOT_FALLBACK enables Playwright-based individual section screenshots for sections outside the initial screenshot range (WebGL/lazy-rendered pages). This significantly improves DINOv2 visual embedding coverage. Set to "false" to disable.

Optional environment variables (defaults work out of the box): MAX_TILES_PER_SECTION (default 20, max 100) -- max tiles per section for multi-tile capture. BLANK_IMAGE_STDDEV_THRESHOLD (default 5.0) -- stddev threshold for blank image detection. DUPLICATE_VECTOR_THRESHOLD (default 0.995) -- cosine similarity threshold for vision embedding dedup. EMBEDDING_IDLE_TIMEOUT_MS (default 30000) -- ONNX Worker VRAM auto-release timer (0 to disable). DINOV2_MODEL_PATH -- custom DINOv2 ViT-B/14 ONNX model path. EMBEDDING_CACHE_ENABLED (default true) -- enable/disable the Layout Embedding disk cache (additive opt-out flag; set "false" to write no cache files). REFTRIX_EMBEDDING_CACHE_ROOT (default /tmp/reftrix-embedding-cache) -- embedding cache root; a root resolving outside os.tmpdir() is rejected by default (fail-closed). Set REFTRIX_EMBEDDING_CACHE_ROOT_ALLOW_FALLBACK=true to instead degrade to the default root with a warning.

Example tools

ReftrixMCP provides 28 MCP tools. Key examples:

  • layout.ingest -- fetch a web page, take a screenshot, and extract section patterns

  • layout.search -- semantic search over layout sections by natural-language query

  • motion.detect -- detect CSS/JS animations with video-mode frame capture

  • quality.evaluate -- score design quality on originality, craftsmanship, and contextuality

  • page.analyze -- unified analysis: layout + motion + quality + responsive in one call (async via BullMQ), with opt-in Phase 7.5: accessibility audit, performance evaluation, and auto snapshot

  • responsive.search -- search responsive analysis results by viewport and breakpoint

  • part.search -- semantic search over UI parts with visual (DINOv2) or text embeddings

  • part.inspect -- get detailed part info including computed styles, bounding box, and accessibility

  • part.compare -- compare 2-5 parts side by side on styles, layout, and interaction

Full tool reference: MCP Tools Guide

Architecture

MCP Client (Claude Desktop / Code)  --stdio-->  MCP Server (<!-- gen:tool-count -->28<!-- /gen:tool-count --> tools, Zod)
  +-- Service Layer: Playwright, Sharp+Pixelmatch, DOMPurify
  +-- ML Layer: ONNX Runtime (multilingual-e5-base + DINOv2 ViT-B/14, 768-dim)
  +-- BullMQ Workers: page.analyze, quality.evaluate
  +-- PostgreSQL 18 + pgvector 0.8 (HNSW, tsvector)  +  Redis 7

Documentation

Guide

Description

Getting Started

Installation, setup, and first analysis

MCP Tools Guide

All 28 tools with usage examples

page.analyze Deep Dive

Async analysis flow and data structures

Troubleshooting

Common issues and solutions

Known limitations

  • onnxruntime-node is an optional dependency that pnpm install installs by default (it powers the ML features — embedding and visual search). If it fails to install on an unsupported platform, or you skip it with pnpm install --no-optional, the non-ML tools (layout analysis, quality evaluation, code generation) still work

  • CPU-mode embedding takes ~2-5 s per text; GPU recommended for batch workloads

  • Minimum 16 GB RAM; 32 GB recommended for concurrent analysis with Ollama Vision

  • First embedding call downloads ~1.1 GB ONNX model (multilingual-e5-base, FP32) into the transformers.js cache. Verify integrity at any time with pnpm --filter @reftrixmcp/ml repair:e5-cache --check; pass --repair to re-download on size/SHA-256 mismatch, or --force to always re-download

  • page.analyze workers are auto-forked by WorkerSupervisor when the MCP server starts (v0.4.0 PR7d-2+); manual start is developer-only (REFTRIX_ALLOW_MANUAL_WORKER=true required when MCP server is running)

  • Vision analysis (layout, motion, narrative) requires Ollama + llama3.2-vision running locally

  • DINOv2 visual embedding model requires ~330 MB download (ViT-B/14 ONNX)

Release notes / リリースノート

  • npm publish automation — Trusted Publishing / OIDC (2026-07-12): npm publishing is now driven by CI. Creating a GitHub Release for a v* tag triggers .github/workflows/publish.yml, which publishes the 5 packages (@reftrixmcp/core, @reftrixmcp/database, @reftrixmcp/ml, @reftrixmcp/webdesign-core, @reftrixmcp/mcp-server) in dependency order via npm Trusted Publishing (OIDC) — no NPM_TOKEN secret is used, and every package is published with --provenance. A verify job builds and validates all tarballs first; a publish job runs only after the npm-publish GitHub Environment's required-reviewer approval. / npm 公開自動化 — Trusted Publishing / OIDC(2026-07-12): npm 公開は CI 駆動になりました。v* タグの GitHub Release を作成すると .github/workflows/publish.yml が起動し、5 パッケージ(@reftrixmcp/core・@reftrixmcp/database・@reftrixmcp/ml・@reftrixmcp/webdesign-core・@reftrixmcp/mcp-server)を依存順に npm Trusted Publishing (OIDC) で公開します — NPM_TOKEN シークレットは使用せず、各パッケージは --provenance 付きで公開されます。verify job が先に全 tarball を build・検証し、publish job は npm-publish GitHub Environment の required-reviewer 承認の後にのみ実行されます。

  • Plan v4.4 PR-N (2026-05-17): WorkerSupervisorOptions.restartDelayMs field formal removal + env-only canonical SSOT consolidation per ADR-0035 Amendment 1 §Decision 5. The WORKER_RESTART_DELAY_MS and EMBEDDING_BACKFILL_RESTART_DELAY_MS environment variables are now the sole source of truth for per-type restart cooldown values; resolution is performed via getRestartDelayMsForType(workerType). Server version bumped to 0.6.0. / WorkerSupervisorOptions.restartDelayMs フィールドを正式削除し、ADR-0035 Amendment 1 §Decision 5 に従い env-only canonical SSOT へ一元化。WORKER_RESTART_DELAY_MS と EMBEDDING_BACKFILL_RESTART_DELAY_MS 環境変数が per-type restart cooldown 値の唯一の真実源となり、getRestartDelayMsForType(workerType) 経由で解決される。サーバーバージョンを 0.6.0 に bump。

License

AGPL-3.0-only -- see LICENSE.

Network use requires source disclosure per Section 13. Source: github.com/TKMD/ReftrixMCP Commercial license: licence@reftrix.io

Contributing

See CONTRIBUTING.md.

Security

Report vulnerabilities per SECURITY.md. Privacy: docs/legal/PRIVACY_POLICY.md | Profiling privacy: apps/mcp-server/PRIVACY.md | Data retention: apps/mcp-server/DATA_RETENTION.md | Third-party licenses: THIRDPARTY_LICENSES.md

Available Tools

27 tools
accessibility.auditA
Read-onlyIdempotent

WCAG 2.1 accessibility audit using axe-core with contrast ratio checking. Analyzes HTML or URL for WCAG A/AA/AAA compliance, detects violations with severity classification, calculates accessibility score (0-100), and checks text/background contrast ratios.

Inputs: url, html, level, include_passes, +1 more.

audit.queryB
Read-onlyIdempotent

監査ログを検索します。GDPR Art.30に基づく処理活動記録の閲覧。Query audit logs. View records of processing activities per GDPR Art.30.

Inputs: limit, action, end_date, start_date, +1 more.

background.searchA
Read-onlyIdempotent

BackgroundDesignをセマンティック検索します。グラデーション、グラスモーフィズム、SVG背景等の背景デザインパターンを自然言語で検索できます。designType(14種類)やwebPageIdでフィルタリング可能です。

Inputs: query (required), limit, offset, filters, +1 more.

data.deleteA

GDPR Art.17「忘れられる権利」/ CCPA §1798.105 に基づくデータ完全削除。page(全関連テーブルCASCADE DELETE)、profile(嗜好プロファイル完全削除)、all_user_data(全ユーザーデータ一括削除)から選択。confirm: true 必須。target=page / all_user_data では DB 削除前に embedding-backfill Queue の滞留ジョブ(7カテゴリ)も削除し、非同期 backfill による削除済みデータ復活を防止する。GDPR Art.17 / CCPA §1798.105 Right to Erasure. Permanently deletes all data for the specified target. Supports page (CASCADE DELETE), profile (hard delete), all_user_data (bulk delete). confirm: true is required. For target=page / all_user_data, embedding-backfill queue jobs (7 categories) are also removed before DB deletion to prevent async backfill from resurrecting erased data.

Inputs: id (required), reason (required), target (required), confirm (required), +1 more.

data.exportA
Read-onlyIdempotent

GDPR Art.20「データポータビリティの権利」に基づくデータエクスポート。指定されたpage/profileの全関連データをJSON形式でエクスポート。PII情報を明示的にマーキング。GDPR Art.20 Right to Data Portability. Exports all related data for the specified target in JSON format. PII fields are explicitly marked.

Inputs: id (required), target (required).

design.compareA
Read-onlyIdempotent

2-5件のWebページをレイアウト・視覚・品質・カラーの4軸で比較し、ペアワイズ類似度スコア(0-1)を算出します。include_detailsで共通パターンと差分ポイントも取得可能。 / Compare 2-5 web pages across layout, visual, quality, and color dimensions. Returns pairwise similarity scores (0-1). Set include_details for common patterns and key differences.

Inputs: page_ids (required), dimensions, include_details.

design.search_by_imageA
Read-onlyIdempotent

画像から視覚的に類似したデザインセクションを検索します。Base64エンコード画像またはHTTPS画像URLを入力として受け付けます。DINOv2 visual embeddingを使用したHNSW検索で類似デザインを発見します。オプションのテキストクエリを指定すると、RRF 3-source融合(text 40% + vision 30% + fulltext 30%)でハイブリッド検索を実行します。

Inputs: image (required), limit, query, section_type, +1 more.

design.similar_siteA
Read-onlyIdempotent

URLを入力として、DB内の類似デザインのWebサイトを検索します。指定URLのページのセクションembedding(DINOv2 vision + e5-base text)のmean poolingでページレベルの代表ベクトルを生成し、pgvector HNSW検索で類似サイトを発見します。RRF 3-source fusion(text 40% + vision 30% + fulltext 30%)で総合スコアを算出。 / Searches for similar website designs in DB given a URL. Generates page-level representative vectors via mean pooling of section embeddings (DINOv2 vision + e5-base text) and finds similar sites using pgvector HNSW search.

Inputs: url (required), limit, include_details.

layout.generate_codeA
Idempotent

セクションパターンからReact/Vue/Svelte/Astro/HTMLコードを生成します。パターンIDを指定して、選択したフレームワーク(React, Vue, Svelte, Astro, HTML)でコードを出力できます。TypeScript/JavaScript、Tailwind CSS/Vanilla CSSの選択も可能です。

Inputs: patternId (required), options.

layout.ingestA

Fetch HTML/screenshot from URL for layout analysis. SSRF protection blocks private IPs/metadata services. HTML is sanitized.

Inputs: url (required), options, source_type, usage_scope.

layout.inspectB
Read-onlyIdempotent

Parse HTML and extract section structure, grid, typography info

Inputs: id, html, options.

layout.searchB
Read-onlyIdempotent

セクションパターンを自然言語クエリでセマンティック検索します。日本語・英語の両方に対応しています。hero、feature、cta、testimonial、pricing、footer等のセクションタイプでフィルタリングできます。use_vision_search=trueでvision_embeddingを使用したハイブリッド検索(RRF: 60% vision + 40% text)が可能です。

Inputs: query (required), limit, offset, filters, +12 more.

motion.detectC
Read-onlyIdempotent

Detect/classify motion patterns from web page. Parses CSS animations, transitions, keyframes. Warns about performance/accessibility issues.

Inputs: css, url, html, pageId, +15 more.

motion.searchA
Read-onlyIdempotent

モーションパターンを類似検索、または実装コードを生成します。action: search(デフォルト)で検索、action: generateでCSS/JS実装コードを生成します。

Inputs: limit, query, action, format, +11 more.

narrative.searchB
Read-onlyIdempotent

世界観・レイアウト構成でセマンティック検索します。自然言語クエリ(例: "サイバーセキュリティ感のあるダークなデザイン")または768次元Embeddingで検索可能。Hybrid Search(Vector + Full-text)でRRF統合。

Inputs: query, filters, options, embedding, +1 more.

page.analyzeA
Read-onlyIdempotent

Analyze a web page URL with layout detection, motion pattern extraction, and quality evaluation. Executes layout.ingest, motion.detect, and quality.evaluate in parallel and returns unified results. Supports MCP streaming progress via _meta.progressToken for real-time phase notifications.

Inputs: url (required), async, summary, timeout, +23 more.

page.batch_analyzeA

Batch analyze multiple URLs in parallel. Submits all URLs as async BullMQ jobs and returns immediately with a batch ID.

Features:

  • Up to 50 URLs per batch

  • Configurable concurrency (1-5, default: 3)

  • SSRF validation for all URLs before submission

  • Graceful degradation: individual failures don't affect the batch

  • Progress tracking via page.getJobStatus(batch_id)

  • Rate limiting within analysis tier (10 RPM)

Use page.getJobStatus with batch_id to poll for results.

Inputs: urls (required), timeout, features, on_error, +3 more.

page.getJobStatusA
Read-onlyIdempotent

Check the status of an async page analysis job or of a batch.

Pass exactly one of:

  • job_id: the job ID returned by page.analyze(async=true)

  • batch_id: the batch ID returned by page.batch_analyze

Returns for job_id:

  • Job state (waiting, active, completed, failed)

  • Progress percentage (0-100)

  • Result summary when completed

  • Error details when failed

Returns for batch_id:

  • Batch state (waiting, active, completed, failed, partial)

  • Progress percentage (0-100)

  • Summary (total, completed, failed, skipped counts)

  • Individual job results with states and webPageIds

Note: Requires Redis to be running. Jobs are retained for 24 hours after completion; batches for 48 hours.

Inputs: job_id, batch_id.

part.compareB
Read-onlyIdempotent

2-5個のUIパーツをスタイル・レイアウト・インタラクション・アクセシビリティで並列比較。デフォルトはstyles+layout。各プロパティの同一性も判定。 / Compare 2-5 UI parts on styles, layout, interaction, and accessibility. Default: styles + layout. Reports property-level identity.

Inputs: part_ids (required), compare_aspects.

part.inspectA
Read-onlyIdempotent

特定のUIコンポーネントパーツの詳細情報を取得します。スタイル、HTML、バウンディングボックス、インタラクション情報、Embedding有無等を返します。 / Inspect a specific UI component part by ID. Returns styles, HTML, bounding box, interaction info, embedding status, etc.

Inputs: part_id (required), include_html, include_embedding.

part.searchB
Read-onlyIdempotent

UIコンポーネントパーツ(ボタン、カード、リンク等)をセマンティック検索します。テキストクエリ(e5-base + full-text)によるハイブリッド検索を提供。partType(16種類)やsearchMode(visual/text/hybrid)でフィルタリング可能です。 / Search UI component parts (buttons, cards, links, etc.) by text query. Provides hybrid search via e5-base + full-text. Filterable by partType (16 types) and searchMode (visual/text/hybrid).

Inputs: tags, limit, query, offset, +7 more.

performance.evaluateA
Read-onlyIdempotent

Evaluate web page performance using Core Web Vitals (LCP, FID, CLS, INP, TTFB) via Playwright PerformanceObserver API. Returns a score (0-100), grade, and optional improvement recommendations.

Inputs: url (required), budget, include_details.

quality.evaluateC
Read-onlyIdempotent

Evaluate web design quality on 3 axes (originality, craftsmanship, contextuality) with AI cliche detection

Inputs: html, pageId, strict, context, +8 more.

responsive.captureA
Read-onlyIdempotent

3ビューポート(desktop 1920x1080, tablet 768x1024, mobile 375x812)でWebページを同時キャプチャし、レスポンシブレイアウトの差分を分析します。セクション表示/非表示、フォントサイズ変化、グリッドカラム変化、スペーシング変化を検出し、差分スコア(0-100)を返します。

Inputs: url (required), viewports, include_diff, include_screenshots.

responsive.searchA
Read-onlyIdempotent

レスポンシブデザイン分析結果をセマンティック検索します。ビューポート間の差異(レイアウト変化、ナビゲーション変化、表示切替等)を自然言語で検索できます。差異カテゴリ、ビューポートペア、ブレークポイント範囲、スクリーンショット差分率でフィルタリング可能です。

Inputs: query (required), limit, offset, filters, +1 more.

search.unifiedA
Read-onlyIdempotent

Layout(セクション)・Part(UIコンポーネント)・Motion(アニメーション)・Background(背景デザイン)・Narrative(世界観)を横断的にセマンティック検索します。個別検索ツールを並列実行し、結果をsimilarityスコア順にマージして返却します。 / Cross-component semantic search across Layout sections, UI Parts, Motion patterns, Background designs, and Narratives. Executes individual search tools in parallel and merges results by similarity score.

Inputs: query (required), tags, limit, types, +8 more.

system.healthA
Read-onlyIdempotent

Run MCP server health check. Checks MCP tool metrics, embedding cache stats, service initialization status, pattern services health, and returns diagnostics.

Inputs: include_metrics, include_cache_stats, include_tools_status, include_vision_hardware, +3 more.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    B3.4/5.0

    Scored across 27 tools

    Disambiguation3/5

    Most tools map to distinct resource+action pairs, but there is substantial overlap: search.unified duplicates the seven individual *.search tools, design.search_by_image overlaps design.similar_site, and page.analyze re-runs layout.ingest, motion.detect and quality.evaluate. Descriptions help disambiguate, but an agent must decide between the orchestrator, the batch/async path, and the primitive tools, which creates real misselection risk.

    Naming Consistency4/5

    Almost all tools use a consistent namespace.verb_noun pattern (layout.ingest, motion.detect, part.compare, data.delete). The main deviation is page.getJobStatus, which uses camelCase while siblings like page.batch_analyze use snake_case, breaking the otherwise predictable convention.

    Tool Count2/5

    At 27 tools the surface is heavy for a web-design analysis server, and several tools are redundant (search.unified subsumes the per-resource searches, page.analyze subsumes the analysis primitives). It exceeds the 25-tool threshold where each tool no longer clearly earns its place.

    Completeness4/5

    The domain (web design analysis) is well covered: ingestion, layout/motion/responsive detection, semantic search across sections/parts/backgrounds/narratives, code generation, quality/accessibility/performance evaluation, comparison, and GDPR export/delete with audit logging. Minor gaps exist around pattern update/mutation and there is redundancy rather than missing lifecycle coverage.

    Maintenance

    ActivityNo data
    ResponsivenessOutdated data

    Related MCP Connectors

    Related MCP Servers

    • -
      license
      Not graded
      quality
      F
      maintenance
      This server provides cloud browser automation capabilities using Browserbase, Puppeteer, and Stagehand. This server enables LLMs to interact with web pages, take screenshots, and execute JavaScript in a cloud browser environment.
      1,158 npm
      3,397
      MIT
    • A
      license
      A
      quality
      B
      maintenance
      Firecrawl, the context API for AI agents, as an MCP server. Search the web, scrape any URL, parse PDFs and documents, crawl and map sites, interact with pages in a real browser, run Agent research jobs, monitor pages for changes, and search scientific papers. Returns clean Markdown or structured JSON. Run locally with npx firecrawl-mcp or connect to https://mcp.firecrawl.dev/v2/mcp
      28
      201,139 npm
      7,816
      MIT
    • A
      license
      A
      quality
      D
      maintenance
      A comprehensive MCP server providing 15 web tools including search, scraping, screenshots, SEO audits, and DNS/SSL checks through a single installation. It delivers clean, LLM-optimized outputs so AI agents can focus on reasoning rather than parsing raw HTML.
      15
      17 npm
      MIT