apple-docs-mcp

Executes semantic search queries against Apple's official developer documentation API, returning ranked results with title, summary, and direct documentation links. Implements LRU caching with 10-minute TTL for search results (200 entry limit) to reduce redundant API calls while keeping results fresh for dynamic user queries. Integrates directly with Apple's search infrastructure rather than building a custom index, ensuring compatibility with the latest documentation updates.

Fetches full documentation content for a specific Apple framework, class, or API by URL or identifier, parsing Apple's JSON API responses to extract structured content including method signatures, parameters, return types, and code examples. Implements 30-minute LRU cache (500 entries) for API documentation to optimize repeated lookups of the same framework while respecting Apple's documentation update cadence. Handles both Swift and Objective-C documentation formats transparently.

Organizes WWDC video index by year (2014-2025) enabling developers to filter videos by specific WWDC events or year ranges. Supports queries like 'show me all WWDC 2023 sessions on SwiftUI' or 'find videos from the last 3 years about App Services'. Maintains historical context of how Apple's frameworks and best practices have evolved across WWDC events.

Implements MCP server initialization, configuration loading, and graceful shutdown. Handles TypeScript compilation, environment variable loading, and MCP protocol handshake with clients (Claude Desktop, Cursor, VS Code). Manages server state including cache initialization and tool registry setup. Supports configuration via environment variables and config files.

Retrieves and caches method signatures, parameter types, return types, and availability information from Apple's documentation API. Enables AI assistants to understand the exact signature of an API before generating code that uses it. Validates parameter types and counts to catch potential errors early.

Analyzes user queries to infer intent and recommend relevant documentation, frameworks, or WWDC videos. Uses keyword matching and topic correlation to suggest related documentation that may be useful. For example, a query about 'state management' might recommend SwiftUI documentation, Combine framework docs, and related WWDC sessions.

Supports querying multiple documentation items in a single request and aggregating results. Enables developers to retrieve documentation for multiple APIs, frameworks, or WWDC videos in parallel, reducing round-trip latency. Results are aggregated and deduplicated before returning to the client.

Searches a locally-maintained JSON index of 2,000+ WWDC videos (2014-2025) organized across 17 topic categories (SwiftUI, App Services, Developer Tools, Machine Learning, etc.) and chronologically by year. Implements instant local search without external API calls by maintaining an in-memory index of video metadata (title, description, year, topics, video ID). Supports multi-dimensional filtering: by topic (e.g., 'SwiftUI & UI Frameworks'), by year range, and by keyword matching against titles and descriptions.

Retrieves and caches metadata about Apple frameworks (UIKit, SwiftUI, Core ML, etc.) including minimum iOS/macOS version requirements, deprecation status, and platform availability (iOS, macOS, watchOS, tvOS). Implements 1-hour LRU cache (100 entries) for framework indexes since API structure changes infrequently. Enables developers to understand version constraints before using an API in their codebase.

Exposes 15 specialized MCP tools through a standardized schema registry that enables AI assistants (Claude, Cursor, etc.) to invoke Apple documentation functions with type-safe parameters. Each tool is defined with JSON Schema including input parameters, return types, and descriptions. The MCP server handles tool invocation routing, parameter validation, and response serialization according to the Model Context Protocol specification.

Implements a multi-tier LRU cache (cache.ts) with content-type-specific time-to-live values: API documentation (30 min, 500 entries), search results (10 min, 200 entries), framework indexes (1 hour, 100 entries), technology lists (2 hours, 50 entries), and WWDC video data (2 hours, unlimited). Each cache tier is optimized for the update frequency and size of its content type. Cache eviction uses least-recently-used policy when capacity is reached.

Tracks and retrieves framework availability across 4 Apple platforms (iOS, macOS, watchOS, tvOS) with version constraints for each. Queries Apple's API metadata to determine which frameworks are available on which platforms and their minimum version requirements. Enables developers to understand cross-platform compatibility at a glance.

Maintains a curated JSON index of WWDC videos with rich metadata including title, description, year, topic tags (17 categories), and Apple video URL. Each video is tagged with one or more topics (e.g., 'SwiftUI & UI Frameworks', 'App Services', 'Developer Tools') enabling multi-dimensional filtering. Topic taxonomy is manually maintained by the MCP project and covers major Apple development areas.

Transparently handles both Swift and Objective-C API documentation formats from Apple's documentation API, normalizing differences in method signatures, parameter naming conventions, and return types. Presents unified documentation regardless of whether the underlying API is Swift-native, Objective-C-native, or bridged. Enables developers to work with either language without needing separate documentation lookups.

Extracts code examples from Apple's documentation API responses and presents them in context with surrounding documentation. Identifies code snippets in documentation content and formats them for readability in AI assistant interfaces. Supports both Swift and Objective-C code examples with syntax preservation.