SMRY

SMRY is an AI-assisted news and RSS reader that cleans articles, summarizes them with traceable passages, files saves automatically and answers questions across a searchable personal library, with an API and MCP server for agents.

SMRY is a reading workspace that combines a news and RSS reader with a personal library and an AI assistant. Pages, videos, PDFs and feed items are stripped down to clean text, summarized, tagged and made searchable, so that material read months ago can still be found and questioned. The pitch is a single place for following sources, reading without clutter, and retaining what was read. The feed side covers RSS and Atom, public newsletter editions, podcasts, YouTube channels, Reddit communities, public X posts, and selected webpages turned into monitored sources. Items land in a finite queue that can be organized with collections and saved views, and natural-language filters keep or drop matching posts. Subscriptions move in and out through OPML. The reader side offers adjustable fonts, themes and text size, audio playback with word-level highlighting, five-color highlights with notes, and export to tools such as Notion, Obsidian, Anki and Zotero. The assistant, named Remy, answers questions across the whole library and points back to the exact passages used. Summaries show which paragraphs they came from, and answers separate source claims from inference. Study features include quizzes, flashcards and idea maps, plus a signal that estimates whether a text was written by a human or an AI. Rule-based auto-tagging files new saves without manual sorting. SMRY is aimed at heavy readers, researchers, students and people who follow many publications and want fewer tabs and apps. It runs as an offline-capable web app with extensions for Chrome, Firefox, Edge and Opera, and Safari is still awaiting release. Every signed-in account also gets a Developer API and an MCP server, which lets agents search the same library and feeds. Compared with plain RSS readers or read-later apps, it adds source-grounded AI; compared with general chatbots, it keeps a persistent, organized archive of what the user has actually read.