Apple RAG MCP MCP

The Apple docs MCP your AI actually deserves.

Apple docs. WWDC26 transcripts. RAG retrieval + Jev relevance ranking. One clean tool.

Website · Powered by Jev

· English | 中文

Not Just Another Docs Tool

Hybrid retrieval combines keyword and semantic search to find Apple sources. Jev then ranks the merged candidates for relevance to your question, bringing the best matches to your agent.

Minimal footprint. Maximum signal. Our MCP tools are designed to be lean—no bloated responses, no wasted tokens, no noise cluttering your agent's context. Just the information that matters.

Powered by Jev

Jev, a decision model, scores candidate Apple documentation and video transcripts against each query's API, platform, and version requirements. We use its relevance scores to rank the sources returned through MCP. A backup reranker takes over if the API call fails.

Open Source. Ready to Use.

The hosted service connects your agent to an Apple knowledge base that is already collected, cleaned, and indexed.

• Ready-to-query sources: Official developer documentation and video transcripts, available through one connection.
• Retrieval model calls included: Embeddings, Jev ranking, and the backup reranker run on our service.
• Ongoing upkeep: We handle content collection, index updates, database backups, and protocol compatibility.

The source is public for inspection and self-hosting. Running your own instance means operating a database, building and refreshing your corpus, and managing your model API credentials.

Start free. Pro includes 50,000 tool calls/week and 50/minute for $1/week, shared by search and fetch.

Start in Seconds

Configure with your agent: In the dashboard, select a token and click Copy Install Prompt. Paste it into your current agent to configure the connection and verify the tools.

One click:

Open Cursor and confirm installation.

Option 2: Manual Setup for Other MCP Clients

JSON Configuration (Copy & Paste): This is a configuration example, not a universal client format. The dashboard also retains manual configuration examples.

{
  "mcpServers": {
    "apple-rag-mcp": {
      "type": "http",
      "url": "https://mcp.apple-rag.com"
    }
  }
}

Manual Configuration Parameters:

• MCP Type: Streamable HTTP
• URL: https://mcp.apple-rag.com
• Protocol: 2026-07-28
• Authentication: Optional (MCP Token for higher limits)
• MCP Token: Get yours at apple-rag.com for increased quota

Client compatibility requires support for protocol 2026-07-28; saving a configuration alone does not verify a working connection.

Note: No MCP Token required to start! You get free queries without any authentication. Add an MCP Token later for higher usage limits.

🌟 Why Developers Love Apple RAG MCP

🎯 Features

• 🔍 Semantic Search - Vector similarity with semantic understanding for intelligent retrieval
• 🔎 Keyword Search - Precise technical term matching for API names and specific terminology
• 🎯 Hybrid Search - Combines keyword and semantic retrieval
• 📚 Complete Coverage - iOS 27, iPadOS 27, macOS 27, watchOS 27, tvOS 27, and visionOS 27 documentation
• 🎬 WWDC26 Videos - Full transcripts from Apple Developer videos and WWDC26 sessions
• ⚡ Fast Response - Optimized for speed across all content types
• 🚀 High Performance - Multi-instance cluster deployment for maximum throughput
• 🔄 Always Current - Synced with Apple's latest docs, WWDC26 sessions, and Xcode 27 beta references
• 🛡️ Secure & Private - Your queries stay private
• 🌐 MCP 2026 Native - Stateless MCP 2026-07-28

📄 License

This project is licensed under the MIT License.