What problem does it solve? LLMs write SwiftUI from stale training memory, producing deprecated modifiers, hallucinated argument shapes, and iOS-isms that don't fit macOS apps. This Skill replaces memory with evidence by querying a corpus of 1,857 real open-source macOS apps parsed with SwiftSyntax, so every API call is grounded in how shipping apps actually write it. ## Core Features & Use Cases - API Lookup with Consensus Shapes: Run swiftui-ctx lookup <api> to see the argument shapes real apps use (with percentages), a quality-ranked recommended example, deprecation status, and a link to the official sosumi.ai docs. - Real Source Retrieval: Use swiftui-ctx file <id> --smart to fetch the actual compilable enclosing view from GitHub, pinned to a commit permalink, instead of emitting one-line fragments. - Recipes and Intent Search: Find multi-API production patterns (menubar-app, settings-screen, nsview-bridge) via recipe, or discover which APIs to use from a plain-language intent via search. - Use Case: Before writing a search field, run swiftui-ctx lookup searchable, read the consensus shape, then file --smart the recommended example to write the modifier exactly as high-quality Mac apps do. ## Quick Start Ask the agent to look up how production macOS apps use the SwiftUI searchable modifier and write the code based on the recommended real-world example.
