# PickFlick > PickFlick is a trailer-first movie & TV discovery app: swipe trailers to build a taste > profile, get recommendations, and share picks with people you know. ## Full application description `/llms-full.txt` explains the whole product — trailer-first discovery, the three-tier swipe vote, watch verdicts, rooms and matches, the library, taste-driven recommendations, and sharing — written from the code, in one page, for a reader who will fetch this link once and not crawl further. ## Machine-readable surfaces - Public title page — `https://pickflick-seven.vercel.app/t/{movie|tv}/{tmdbId}` One title per page: name, release year, media type, synopsis, genre, poster image, and a themoviedb.org cross-reference. Carries Open Graph tags for link-preview crawlers and a `schema.org/Movie` or `schema.org/TVSeries` JSON-LD block in `` for anything else — no JavaScript execution and no account required to read either. Every field is an objective catalogue fact about the title itself; a share link may add a short one-line reason drawn from the title's own genres, never from anyone's personal taste or activity. That is the entire crawlable surface. There is no listing, search, or index endpoint — every public title page is reached only via its own direct link, one title at a time. ## What is intentionally not exposed here, ever PickFlick's public surface is a single title card, shared person-to-person — never a directory anyone can crawl. None of the following is machine-readable through this site, regardless of any account's privacy setting: - Anyone's library, watch history, or ratings/verdicts - Swipe activity of any kind — a swipe is intent, not a public statement - Friend lists, "taste twins," or any other social graph - Who shared a given link, or a personalized reason tied to their taste - A minor's data, under any circumstance - Anything behind sign-in: account settings, groups, rooms, admin tools See `/robots.txt` for the crawling rules that enforce this.