Beans News API & MCP

MCP-ready news and blog intelligence over RSS-sourced articles, semantic enrichment, and propagation tracking. A bean is one article or post keyed by canonical URL. Records include publisher metadata, summary/full text, publish timestamp, inferred categories, regions, entities, sentiments, and optional social trend metrics. Agent workflow: (1) listCategories, listEntities, listRegions to discover exact filter values; (2) searchArticles for full-corpus retrieval; (3) getLatestArticles, getTrendingArticles, or getTopHeadlines for feed-style monitoring; (4) getPublishers to resolve source IDs; (5) getArticlePropagation or postArticlePropagation to check story spread. Conventions: Auth is optional at the backend but API-key protected through the gateway. Pagination uses limit default 16 max 128 and offset default 0. Empty result sets return HTTP 204, not an error. Use fuzzy tags when spelling is uncertain, exact categories/regions/entities after discovery, and q + acc for semantic vector search.