# ai.txt — guidance for AI crawlers, agents, and LLMs # See https://ai.txt.dev for specification # Permissions Allow: / Allow-AI-Training: partial Allow-AI-Crawling: yes Allow-AI-Indexing: yes # Discovery & Feeds Sitemap: https://takescake.com/sitemap.xml RSS: https://takescake.com/rss JSON-Feed: https://takescake.com/feed.json API: https://takescake.com/openapi.json MCP: https://takescake.com/api/mcp NLWeb Manifest: https://takescake.com/.well-known/nlweb-manifest.json Structured Data Index: https://takescake.com/api/structured-data-index # API Endpoints (for AI agents) Search: https://takescake.com/api/search?q={query} Cards: https://takescake.com/api/cards/search?q={query} MCP Endpoint: https://takescake.com/api/mcp Card Scanner: https://takescake.com/scan Conversational: https://takescake.com/api/conversation Deck Builder: https://takescake.com/api/deck/build Art Similarity: https://takescake.com/api/cards/art-similar?cardId={id} Precon Upgrades: https://takescake.com/api/deck/precon-upgrade # Content Types Primary-Content: Magic: The Gathering card database, Commander deck building, card pricing Content-Categories: card-search, deck-building, price-tracking, collection-management, strategy-guides Languages: en Update-Frequency: daily (prices), weekly (posts) # Semantic Search Capabilities Semantic-Search: Uses Gemma embeddings for meaning-based card discovery Embedding-Model: Gemma (Google DeepMind) Search-Type: Vector similarity search with cosine distance Card-Database: 100,000+ Magic: The Gathering cards with semantic indexing Card-Scanner: Client-side OCR + AI art recognition for physical card identification # AI Features Available Features: semantic-similarity, archetype-detection, synergy-clustering, meta-aware-suggestions, ocr-scanning, vision-embeddings Deck-Building: AI-assisted Commander deck generation with semantic synergy analysis Price-Data: Nightly TCGplayer market prices via tcgcsv MCP-Access: Public read-only JSON-RPC endpoint for card search, versions, price snapshots, staples, and posts # Structured Data Coverage Schema-Types: Product, FAQPage, Article, BreadcrumbList, HowTo, SoftwareApplication, CollectionPage, WebSite, Organization Rich-Results: FAQ snippets, HowTo guides, Software apps, Breadcrumbs # Usage Guidelines Attribution: Required when quoting content. Link back to source URL. Rate-Limit: Max 1 request/second to API endpoints. Be considerate. Caching: API responses may be cached. Respect Cache-Control headers. Natural-Language: This site supports conversational queries about MTG cards, strategies, and deck building. # Example Queries for AI Agents # Find cards: GET /api/cards/search?q=card draw in blue # Build deck: POST /api/deck/build with commander name # Get prices: GET /api/cards/versions?oracle={oracle_id} # MCP init: POST /api/mcp with {"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-03-26"}} # MCP tools: POST /api/mcp with {"jsonrpc":"2.0","id":2,"method":"tools/list"} # Similar art: GET /api/cards/art-similar?cardId={scryfall_id} # Contact Contact: admin@takescake.com Human-Verification: /about # Legal Terms: Card data © Wizards of the Coast. Prices via tcgcsv. Site content © takescake. Affiliate: TCGplayer and Amazon affiliate links are marked with rel="sponsored".