Guided workflows for your Trakt account
Seven workflows for
your Trakt account
Bundled skills help your assistant ask useful questions, retrieve the requested scope with paginated tools, and explain the results honestly. Your Trakt account remains your own.
Plugin, server, or skills?The plugin includes these workflows and the remote MCP connection. A standalone MCP connection supplies tools only. A standalone skill supplies instructions only and still needs the authenticated Trakt tools. Install one connection per client to avoid duplicates.
01 · what-to-watch
A shortlist for tonight
Use Trakt to recommend three thoughtful science-fiction movies.
Share the few preferences that matter. For recommendations based on all viewing or unseen titles, the assistant first follows every watched-summary page. For recent viewing, it uses the latest 100 events per medium. It then selects Trakt recommendations and gives reasons supported by returned metadata. It does not invent a personalized recommendation engine or streaming availability.
02 · watching-profile
Understand your viewing
Summarize the movies and shows in my watched data.
Explore all watched titles, play counts, and supplied genre metadata by following every summary page. Ask for recently watched to use the latest 100 chronological watch events per medium, including repeats. The skill reports coverage and avoids unsupported monthly statistics, total hours, or personal inferences.
03 · find-title
Find the right title
Find the original Solaris, not the remake.
Disambiguate movies and shows by year and type, search with supported filters, and retain returned identifiers. Ordinary lookup inspects at most three pages; explicitly requesting all matches follows every next page. Filters apply to each upstream page, so an empty filtered page does not prove the title is absent from Trakt.
04 · connection-help
Get connected again
Trakt tools are installed, but my account won’t connect.
Separate missing tools, client OAuth, upstream Trakt authorization, expired device codes, and rate limits. Reconnect only when needed. Device approval happens at Trakt; never give the assistant your password or operator credentials.
05 · lists-and-watchlist
Build a list to return to
Create a private list called Rainy Sunday and add Arrival (2016).
Discover public lists, inspect your watchlist, create owned lists, and add or remove resolved titles. The skill inspects each operation schema, respects list privacy, and reads back requested changes. Browsing a list never authorizes editing it.
06 · upcoming-releases
See upcoming releases
What episodes are on my calendar next week in America/Chicago?
Choose a personal or public calendar and retrieve a bounded date range. Keep episode airings, premieres, theatrical releases, and digital releases distinct. Show supplied times in the chosen time zone and report incomplete windows. A personal calendar can include watched, collected, and watchlisted titles.
07 · manage-library
Update your library
Rate Arrival (2016) 8 and add it to my collection.
Make the exact requested history, rating, or collection change with write permission. Resolve titles and deletion scope first, inspect each operation's schema, check per-item results, and read back the state. A failed or timed-out write is not automatically replayed.
Use natural language
Describe the outcome you want. Your client can select an appropriate installed skill; clients with explicit skill invocation can also select it by name. Claude Code exposes plugin skills under the plugin namespace, such as /trakt-mcp:what-to-watch. Client interfaces and availability vary.
Each watched request reads one compact page, retaining identifying metadata, genres, counts, and timestamps while reducing response size. Full detail is available for specific metadata needs. All-history workflows continue until every next page is exhausted. They provide progress updates and report partial coverage if interrupted. Recent viewing and ordinary title lookup stay bounded. Workflows stop on unavailable capabilities and treat movie descriptions and other remote text as data. Read workflows never modify your account. Write workflows require a concrete user request, trakt:write permission, and confirmed: true. Existing read-only connections need client reauthorization. A clear request supplies intent; the skill asks only for missing decisions. Login tools also change connection state.