vozclara
Generate Knowledge Packs — summary, flashcards, quiz, transcript — from any YouTube video. Plus OAuth-gated library search, Q\&A, and Anki export.
README
VozClara
Multilingual AI study tool. Paste any public video URL → get a structured Knowledge Pack (summary, key ideas, glossary, quiz, transcript with timestamps) in your native language: Spanish, Portuguese, German, English.
Live: vozclara.app MCP Server: smithery.ai/server/salvador7eon/vozclara
Klassisch in der Haltung, modern im Werkzeug.
MCP integration
VozClara exposes a Model Context Protocol (MCP) server so AI agents (Claude Desktop, Cursor, Claude Code, Continue, VS Code with Copilot Chat, etc.) can call it directly.
One-click install via Smithery → choose your client → done.
Manual config for Claude Desktop (%APPDATA%\Claude\claude_desktop_config.json on Windows,
~/Library/Application Support/Claude/claude_desktop_config.json on macOS):
{
"mcpServers": {
"vozclara": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://vozclara.app/api/mcp"]
}
}
}
Restart Claude Desktop and ask:
"Use vozclara to summarise this video in German: https://www.youtube.com/watch?v=…"
Available tools
The anonymous endpoint (/api/mcp) exposes one tool that works without
sign-in. The OAuth-protected endpoint (/api/mcp/pro) adds three tools
that operate on the user's own VozClara library.
| Tool | Endpoint | Inputs | Output |
|---|---|---|---|
vozclara_generate_pack |
/api/mcp (anonymous) |
url, language (es/pt/de/en), depth (short/standard/deep) |
Knowledge Pack text + structured metadata + deep-link to the interactive pack on vozclara.app |
vozclara_search_my_library |
/api/mcp/pro (OAuth) |
query, optional limit (1-10) |
Semantic-search hits across the authenticated user's packs, each linked to vozclara.app |
vozclara_ask_video |
/api/mcp/pro (OAuth) |
pack_id, question |
RAG-grounded answer with inline citations, strictly from that one pack |
vozclara_export_anki |
/api/mcp/pro (OAuth) |
pack_id |
Deep-link URL the user opens to download the .apkg deck |
Authenticated config (Claude Desktop, paid-tier tools — adds the three library-scoped tools above):
{
"mcpServers": {
"vozclara-pro": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://vozclara.app/api/mcp/pro"]
}
}
}
On first call mcp-remote opens the browser, you sign in with the
magic-link, approve the scopes (library:read, library:write,
profile), and the refresh-token is cached locally — same flow as
GitHub's MCP server.
Claude Skills bundle
A higher-level companion to the raw MCP tools above. skills/ in this
repo holds three installable Claude Skills that route everyday video /
knowledge / study asks through the VozClara MCP server — so Claude
knows when to reach for vozclara_generate_pack vs
vozclara_search_my_library vs vozclara_ask_video without the user
having to remember tool names.
mkdir -p ~/.claude/skills
cp -R skills/vozclara-* ~/.claude/skills/
Restart Claude Desktop (or /reload-skills in Claude Code). The
skills appear in the picker. See skills/README.md for the trigger
shapes per skill + the OAuth flow for the two paid-tier ones.
Stack
| Layer | Technology |
|---|---|
| Frontend | Vite + React 18 + TypeScript + Tailwind on Cloudflare Pages |
| Worker | Cloudflare Workers + Durable Objects + KV + D1 + Vectorize |
| LLM (free / Pro) | Workers AI · @cf/meta/llama-3.3-70b-instruct-fp8-fast |
| LLM (Pro Plus) | Anthropic Claude Sonnet 4.5 via Cloudflare AI Gateway |
| Transcripts | Supadata (Merchant-of-Record, licensed) |
| Embeddings | @cf/baai/bge-base-en-v1.5 for semantic search |
| TTS | OpenAI tts-1 (optional, premium narration) |
| Auth | Magic-link via Resend, no passwords |
| Payments | Paddle (Merchant-of-Record, EU VAT + US sales tax handled) |
| MCP | agents SDK + McpAgent + (Phase 2) workers-oauth-provider |
| Analytics | Cloudflare Web Analytics (cookieless, no consent banner) |
Local development
Prerequisites: Node 20+, Git.
# Install
npm install
npm --prefix worker install
# Two terminals — Vite dev server + Wrangler dev
# Terminal A: Vite on :5173
npm run dev
# Terminal B: Wrangler on :8787
npm run worker:dev
Vite proxies /api/* to :8787, so http://localhost:5173 works
end-to-end.
For the worker to call Supadata locally, drop a .dev.vars file inside
worker/:
SUPADATA_API_KEY=sd_…
# Optional — enables OpenAI TTS path
OPENAI_API_KEY=sk-…
# Optional — enables Sonnet 4.5 for Pro Plus tier
ANTHROPIC_API_KEY=sk-ant-…
The file is git-ignored. Production secrets go through
wrangler secret put <NAME> instead.
Deployment
Frontend → Cloudflare Pages
npm run build
# Push to GitHub; Cloudflare Pages builds dist/ automatically.
# Custom domain: vozclara.app
Worker → Cloudflare Workers
cd worker
npx wrangler deploy
This deploys the API + MCP server to vozclara.app/api/* and the
matching Smithery URL at https://vozclara--salvador7eon.run.tools.
Compliance
- DSGVO Art. 13 — full subprocessor list at
/privacy - DSGVO Art. 17 — server-side account deletion sweeps email + sessions + votes + reviews
- EU AI Act Art. 50(1) — AI disclosure banner shown on first generator visit
- EU AI Act Art. 50(2) — every exported pack carries an AI-generated watermark
- Refund policy — 14-day money-back guarantee at
/refund
Brand
LEON MARÉ Editorial:
- Colors — Navy
#0A1A3A, Gold#C9A24B, Creme#F7F3EC, Graphit#1A1A1A - Typography — Per-locale: Reforma (ES), Adelle Sans (PT), Inter (DE), Tiempos (EN)
- Tonalität — classical, declarative, no exclamation marks, no emoji, no superlatives
VozClara · A LEON MARÉ product · Frankfurt am Main · Donostia · Porto
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.