MCP App Store
by KeenEthics
GoodWatch icon

GoodWatch MCP Server for ChatGPT

By GoodWatch

Overview

Choose a movie or show with GoodWatch’s scores, streaming availability and your own taste. Find titles by name or mood, compare details and similar titles, and see what to watch next from your Wishlist and show progress. With your permission, add or remove Want to See titles, rate movies and shows, log watches, update show status and preferences, and create or edit share lists. A GoodWatch account and connection are required. Access is limited to your connected account; read-only access is available. Availability depends on your country, selected services and data freshness. Unknown availability and missing taste matches remain unknown. Smarter descriptive search has a daily allowance and may fall back to name search. Share lists have exactly five titles and are public or accessible to anyone with their link. Video playback happens outside GoodWatch. History imports and account deletion are unavailable through this integration. Destructive changes and publishing require your confirmation.

How to connect

  1. ChatGPT: Open the GoodWatch listing in the ChatGPT app directory and choose Connect. ChatGPT listing GoodWatch exposes 25 tools here.

Tools

create_list

ChatGPT
Create a share list of exactly five distinct titles in rank order. Requires an existing site handle; if missing, returns handle_required and the Lists page. Uses the site's default card and Ember theme. Public by default; unlisted is accessible to anyone with the share link. Returns the site's real share link. Does not change ratings or Wishlist. For title actions, title_id takes precedence; title_name and year are ignored when an ID is supplied. Without title_id, title_name is required and year is optional. Replayed results describe the original action, not current state; a new read returns current state.

discover_titles

ChatGPT
Discover without a query using the same catalog, filters and ordering as GoodWatch Discover/Explore. get_discover_filters describes every filter group, scales, presets, names and example calls. similar_to filters similar titles; more_like/less_like add temporary taste hints; people filters cast/crew; production_countries selects origins, streaming_country selects viewing availability, and rating_country selects age ratings independently. For you by default; everywhere and seen titles included. Choose popularity, newest, score or match ordering. Use on_my_services=true for saved services, or providers for specific services, in your saved country. Tag groups match any pick by default; match_all requires all picks in the named groups. Use seen_only for Seen before, wishlist_only for On your wishlist, minimum_match=60 for 60%+, or outside_usual for taste match 50–59. seen_only and exclude_seen=true are mutually exclusive. Filters apply before pagination. Repeat filters when paging with offset cursors. Returns up to 40 titles and a matching Discover grid page_url.

get_discover_filters

ChatGPT
List Discover metadata for discover_titles and search_titles: exact genres, provider IDs/names in your saved country (plus saved providers), saved services, shared score scales and range presets with existing arguments, Released and length presets, mood keys/names/old-name aliases, taste and list filters, match_all, rating_country and facts_available. Industry/region options include usable production_countries arguments. filter_groups explains every argument with use cases; examples contains valid discover_titles calls. Optional streaming_country scopes provider options without changing saved settings. Reads existing settings without changing them.

get_library

ChatGPT
Your marked titles, including scores, Wishlist / Want to See, watch progress and Not interested. filter=want_to_see selects your Wishlist. Includes added_at (Wishlist date, null when absent), runtime_minutes (movie duration or usual show episode length), and GoodWatch match. Defaults to match (Best match by your taste) when taste is available, otherwise the usual date sort. Use sort=runtime for the longest titles first across your entire filtered library, without per-title lookups. Explicit sorts are preserved. Sorts match My library: added (last added), last (last watched), score (highest first), title (ignores leading articles); waiting puts the oldest Wishlist additions first.

get_list

ChatGPT
Read one of your share lists and its five titles in rank order, with taste match and the site's share link. Other members' lists are not exposed by this tool.

get_my_lists

ChatGPT
Your share lists from Lists, including unlisted lists, newest updated first. Each list has exactly five ranked titles and the share link produced by the site.

get_my_shows

ChatGPT
Only your shows: Continue, Pick up, Waiting for episodes, Want to See, On hold and Dropped.

get_person

ChatGPT
A person's catalog credits with your GoodWatch taste match. Returns the subject’s person link and a link for each credited title. Includes every title, even without a poster or score. Defaults to For you, everywhere, including seen titles. Newest and score sort the whole filmography before paging; exclude_seen uses the site's Not seen yet rule. Links open the person's real GoodWatch page; its grouping can differ.

get_settings

ChatGPT
Read Watching from (country), Show titles in (language), and saved streaming service IDs. Browser language is a saved setting, not a language inferred from your country. No browser locale is supplied by MCP.

get_show_progress

ChatGPT
Your current pass, next episode and paged watches. Dates are unknown unless recorded.

get_similar_titles

ChatGPT
Find more like a movie or show from GoodWatch’s own title-page fingerprint neighbors (up to 80 candidates), with each result’s GoodWatch taste match and posters. Ordered For you by default; for_you=false keeps similarity order. Similarity describes the source title, never your taste match. Searches everywhere and includes seen titles by default; exclude_seen=true applies Not seen yet before paging. Not interested and Dropped are always hidden. kind filters all/movie/show. Keep title and filters unchanged when paging. No paid descriptive search is used.

get_taste

ChatGPT
Your real taste signal and strongest attributes; explains when there is not enough evidence.

get_title

ChatGPT
Catalog details, trailer, availability evidence and your marks. Default include=[cast,scores,analysis] adds five top-billed people and directors/creators, each with a ready-made person link, provider scores on labelled scales, and title analysis (moods, themes, tags, featured tropes). Pass include=[] for the smallest answer, or add seasons/episodes for the website’s season/episode ratings. Episodes use the website’s displayed numbering, which can differ from tracking; page with episode_cursor/episode_limit and optionally season_number. Unknown scores and watch dates stay unknown.

get_tonight

ChatGPT
Choose one thing to watch next, using your saved country and services: a next episode from a show you are watching, otherwise a Wishlist movie or show to start (or a suggestion if the Wishlist is empty). Returns one recommendation rather than the complete Wishlist.

log_watch

ChatGPT
Log one movie or episode watch, an inclusive episode range (season, episode, episode_end), or a whole season (season, whole_season=true). Ranges use the catalog tracking season/episode numbers, mark only aired regular episodes missing in the current pass, never duplicate existing watches, and produce one receipt. Dates remain unknown unless explicitly supplied. Results report every added watch. Batch items are movies, episode ranges or whole seasons; whole-show status transitions are not supported by this tool. To change many titles at once, pass items instead of one title: up to 50 per call. Items are written independently and each reports done, already_set, not_found, ambiguous (with candidates; nothing written) or error, with a summary count and a batch_id; batch_id identifies the complete reversible batch. Full history (hundreds of titles or more) file import is available at: https://goodwatch.app/settings/imports For title actions, title_id takes precedence; title_name and year are ignored when an ID is supplied. Without title_id, title_name is required and year is optional. Replayed results describe the original action, not current state; a new read returns current state.

rate_title

ChatGPT
Set a score from 1 to 10, or clear it with null. Clearing a score is destructive. A movie rating can own one undated watch; show ratings never log episodes or change show status. For a show, season (and episode) score that aired season or episode instead of the whole show: kept apart from the show's score, they never change its score, taste or recommendations. Scoring an episode with no watch in any pass also logs one undated watch of it, which moves the show's status as logging it would; clearing that score removes that watch. A season score logs nothing. The result reports the previous score and exact watch changes. To change many titles at once, pass items instead of one title: up to 50 per call. Items are written independently and each reports done, already_set, not_found, ambiguous (with candidates; nothing written) or error, with a summary count and a batch_id; batch_id identifies the complete reversible batch. Full history (hundreds of titles or more) file import is available at: https://goodwatch.app/settings/imports For title actions, title_id takes precedence; title_name and year are ignored when an ID is supplied. Without title_id, title_name is required and year is optional. Replayed results describe the original action, not current state; a new read returns current state.

search_people

ChatGPT
Search cast and crew by name. Returns canonical person_id values and GoodWatch person links.

search_titles

ChatGPT
Search movie and show names across the GoodWatch catalog, separate from your library. Name lookup by default; for descriptive queries, explicitly set ranked=true for an optional smarter match, ordered For you by default. Defaults to everywhere and includes seen titles. get_discover_filters describes every filter group, scales, presets, names and example calls. similar_to filters similar titles; more_like/less_like add temporary taste hints; people filters cast/crew; production_countries selects origins, streaming_country selects viewing availability, and rating_country selects age ratings independently. Use the shared Discover filters to narrow by type, genres, years, runtime, score, services, moods and more; filters accept catalog genre and provider IDs. Tag groups match any pick by default; match_all requires all picks in the named groups. Repeat filters with offset cursors; ranked cursors inherit their view. Except when paging with a ranked cursor, a missing, empty or whitespace-only query runs the native Discover path, filters and For you default; ranked is ignored and no smarter-match attempt is charged. exclude_seen=true returns only unseen results, filtered before pagination. Returns up to 40 results by default; limit bounds the page size. Different works can share a name and year: use runtime_minutes to distinguish them. Smarter matches allow 20 uncached attempts per member per UTC day, one in flight, normalized queries cached for 60 seconds and paging is free.

set_my_services

ChatGPT
Replace your saved country and streaming services. Country changes and removed services are destructive. Services accept exact names or numeric TMDB IDs; unknown/ambiguous names return choices without writing. Reports the full previous/new country and service IDs. For title actions, title_id takes precedence; title_name and year are ignored when an ID is supplied. Without title_id, title_name is required and year is optional. Replayed results describe the original action, not current state; a new read returns current state.

set_not_interested

ChatGPT
Set or clear Not interested for an unseen, unrated title. Setting it removes an existing Wishlist entry and is destructive. Reports both marks before and after. To change many titles at once, pass items instead of one title: up to 50 per call. Items are written independently and each reports done, already_set, not_found, ambiguous (with candidates; nothing written) or error, with a summary count and a batch_id; batch_id identifies the complete reversible batch. Full history (hundreds of titles or more) file import is available at: https://goodwatch.app/settings/imports For title actions, title_id takes precedence; title_name and year are ignored when an ID is supplied. Without title_id, title_name is required and year is optional. Replayed results describe the original action, not current state; a new read returns current state.

set_settings

ChatGPT
Change your saved country (Watching from) and/or title language (Show titles in). Omitted fields and streaming services are preserved. Country and language are independent. Language browser follows your browser on the site. Reports previous and new preferences; concurrent changes return conflict. For title actions, title_id takes precedence; title_name and year are ignored when an ID is supplied. Without title_id, title_name is required and year is optional. Replayed results describe the original action, not current state; a new read returns current state.

set_show_status

ChatGPT
Apply a native show-status transition. Dropped and bulk Seen are destructive. Seen logs aired episodes with no dates. Watching from Seen starts a new pass while retaining earlier watches. Not started never erases watches. Reports previous/new status, pass and watch changes. To change many titles at once, pass items instead of one title: up to 50 per call. Items are written independently and each reports done, already_set, not_found, ambiguous (with candidates; nothing written) or error, with a summary count and a batch_id; batch_id identifies the complete reversible batch. Full history (hundreds of titles or more) file import is available at: https://goodwatch.app/settings/imports For title actions, title_id takes precedence; title_name and year are ignored when an ID is supplied. Without title_id, title_name is required and year is optional. Replayed results describe the original action, not current state; a new read returns current state.

set_want_to_see

ChatGPT
Add or remove a title from your Wishlist. Removal is destructive. Adding a Seen title means Want to rewatch and preserves watches. Reports previous membership and added-at time. To change many titles at once, pass items instead of one title: up to 50 per call. Items are written independently and each reports done, already_set, not_found, ambiguous (with candidates; nothing written) or error, with a summary count and a batch_id; batch_id identifies the complete reversible batch. Full history (hundreds of titles or more) file import is available at: https://goodwatch.app/settings/imports For title actions, title_id takes precedence; title_name and year are ignored when an ID is supplied. Without title_id, title_name is required and year is optional. Replayed results describe the original action, not current state; a new read returns current state.

undo_watch

ChatGPT
Remove one watch record. Choose a watch_id, or omit it for the most recently logged individual watch of this title. Destructive: results identify the exact removed watch, date, pass and resulting marks; retrying the same action never removes a different watch. Or pass only batch_id (from a batch result) to take back the changes that batch call made, title by title, through the same undo the site uses, except a cleared season or episode score, or a write that wasn't confirmed, which it reports as not_undoable; a title changed again since keeps its newer value, and each title reports undone, partly_undone, already_undone, changed_since, not_undoable or error. A batch is undone once: undoing it again returns the first undo's result. For title actions, title_id takes precedence; title_name and year are ignored when an ID is supplied. Without title_id, title_name is required and year is optional. Replayed results describe the original action, not current state; a new read returns current state.

update_list_titles

ChatGPT
Atomically add and remove titles from one of your share lists. A share list always contains exactly five distinct titles: provide additions and removals together, with equal counts. Remaining titles keep their rank; added titles go at the end in supplied order. Returns previous/new rankings and the site's share link. A concurrent edit returns conflict; read again before choosing a new action. No ratings or Wishlist changes. For title actions, title_id takes precedence; title_name and year are ignored when an ID is supplied. Without title_id, title_name is required and year is optional. Replayed results describe the original action, not current state; a new read returns current state.

Capabilities

InteractiveWrites

Example Prompts

Click any prompt to copy it.

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/goodwatch"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/goodwatch)