{"name":"SMI MCP Server","version":"1.0.0","description":"Semantic search and content discovery for SMI content from Storyblok","endpoints":{"/":"Server information (this page)","/health":"Health check for all services","/sync":"Manual sync trigger (POST)","/stats":"Content statistics by type","/mcp/sse":"MCP Server-Sent Events endpoint","/mcp":"MCP direct protocol endpoint"},"tools":{"hybrid_search_content":"Hybrid search combining keyword matching (slug/title) with AI semantic similarity. Best for both navigational queries and natural language questions.\n\nUse this tool when:\n- Searching for specific sections (e.g., \"taskforces\", \"ncia\") - prioritizes exact slug/title matches\n- Natural language questions (e.g., \"what are SMI task forces?\") - uses semantic understanding\n- You want results that match keywords while maintaining semantic relevance\n\nThis tool boosts results where the query appears in the page slug or title, making it ideal for navigational queries that don't work well with pure semantic search.\n\nSupports pagination via offset and limit parameters.\nSupports field selection via fields parameter to return only specific fields.\n\nReturns: Array of matching content with similarity scores, keyword match indicators, and pagination info.","search_content":"Semantic search across all SMI content from Storyblok. Uses AI embeddings to find content by meaning, not just keywords.\n\nThis tool searches through articles, pages, projects, and all other content types in the SMI library. Use this when you need to find content related to a topic or concept.\n\nSupports pagination via offset and limit parameters for retrieving large result sets in chunks.\n\nSupports field selection via fields parameter to return only specific fields (e.g., [\"id\", \"title\", \"similarity\"]).\n\nReturns: Array of matching content with similarity scores, titles, slugs, metadata, and pagination info.","search_content_by_type":"Semantic search filtered by content type. Use this when you want to search within a specific category of content.\n\nCommon content types include:\n- LayoutNewsArticle: News articles\n- LayoutDefault: Standard pages\n- LayoutProject: Project pages\n- LayoutInsurance: Insurance-related content\n- LayoutTestimonialsVideo: Video testimonials\n\nSupports pagination via offset and limit parameters for retrieving large result sets in chunks.\n\nSupports field selection via fields parameter to return only specific fields (e.g., [\"id\", \"title\", \"similarity\"]).\n\nReturns: Array of matching content of the specified type with similarity scores, titles, slugs, metadata, and pagination info.","get_content_by_slug":"Retrieve specific SMI content by slug or UUID.\n\nUse this when you know the exact slug or ID of the content you want to retrieve. This is faster than semantic search when you have a direct reference.\n\nExamples:\n- Slug: \"news/ai-in-the-built-environment\"\n- ID: \"3b382348-3711-403c-80a1-61bb4fb439a2\"\n\nReturns: The content item with all metadata, or null if not found.","find_related_content":"Find content similar to a given item. Useful for \"related articles\" or \"you might also like\" features.\n\nThis tool uses the embedding similarity of content to find items that are semantically related to a source item. Perfect for discovering connected content or building recommendation features.\n\nSupports pagination via offset and limit parameters for retrieving large result sets in chunks.\n\nSupports field selection via fields parameter to return only specific fields (e.g., [\"id\", \"title\", \"similarity\"]).\n\nReturns: Array of related content with similarity scores and pagination info, excluding the source item itself.","batch_search":"Execute multiple semantic searches in a single request. Efficient for comparing results across different queries or building comparison features.\n\nThis tool processes multiple search queries simultaneously, generating embeddings and executing searches in parallel for improved performance.\n\nUse cases:\n- Compare search results for different keywords\n- Build \"also searched for\" features\n- Execute multiple related searches efficiently\n\nSupports field selection via fields parameter to return only specific fields.\n\nReturns: Object with results for each query, indexed by the original query text.","batch_get_content":"Retrieve multiple content items by their IDs in a single efficient request.\n\nThis tool fetches multiple content items simultaneously, useful for building lists, collections, or retrieving related content in bulk.\n\nUse cases:\n- Fetch multiple specific items for a collection\n- Retrieve content for a list of IDs\n- Build \"featured content\" sections\n- Get full details for search result IDs\n\nSupports field selection via fields parameter to return only specific fields.\n\nReturns: Object with results for each ID, indicating whether each item was found and including the content data."}}