Exposes EarScribe's local-first transcription workflow to Claude and other MCP clients through three resources (local setup guide, FAQ, official links) and two tools (get_local_setup, get_official_links). The server surfaces canonical documentation for running OpenAI Whisper entirely in-browser via WebGPU or WebAssembly, so you can answer questions about privacy-focused transcription without hunting through the website. Useful when you're fielding support questions about on-device AI workflows or need to walk someone through EarScribe's offline model caching and export formats. Read-only, no API keys required, sub-50ms cold start. Runs over stdio and works with Claude Desktop, Cursor, Windsurf, and Continue out of the box.