imdb.com

get-rating

Installation

Adds this website's skill for your agents

browse skills add imdb.com/get-rating-16ss93
Summary

Resolve any IMDb title URL / tt-ID / free-form title reference (movie, TV series, episode, mini-series, short) to its current IMDb rating, total vote count, rating distribution per 1-10 bucket, Metascore, and core title metadata (cast, directors, writers, genres, runtime, certification, plot, languages, countries, poster, canonical URL). Read-only.

SKILL.md
408 lines

IMDb Title Rating Lookup

Purpose

Given an IMDb title URL, IMDb title ID (tt...), or free-form title reference (movie / TV show / TV episode / mini-series / short / documentary), return the current IMDb rating, total vote count, rating distribution (votes per 1-10 bucket when shown), and the core title metadata: primary title, original title (when different), title type (movie / tvSeries / tvEpisode / tvMiniSeries / short / documentary / videoGame), release year (or year range for series), MPAA / TV certification, runtime in minutes, genres, Metascore (when present), top-billed cast with role names, directors, writers, primary poster URL, short + long plot summary, language(s), country/countries of origin, and the canonical IMDb URL. For TV episodes additionally return parent series ID + title and season/episode numbers. Read-only — never click Rate, Add to Watchlist, Sign In, or any mutation control.

When to Use

  • "What's the IMDb rating of {movie/show}?"
  • Bulk enrichment of a watchlist / spreadsheet of titles — pass a free-form name or a known tt-ID per row.
  • Comparing the user-rating + Metascore + distribution shape across a candidate set.
  • Resolving an ambiguous free-form title to a canonical tt-ID before scraping any other IMDb subpage.
  • Pulling the JSON-LD aggregateRating for any IMDb title type, including TV episodes (/title/tt.../episodes/).

Workflow

The optimal flow is two-staged:

  1. Resolve free-form input → tt-ID via IMDb's public-but-undocumented suggestion API (no auth, no anti-bot, no proxy). This is the same JSON the IMDb search-bar typeahead uses. Always use this first unless the caller already passed a tt-ID or a /title/tt.../ URL.
  2. Fetch the canonical title page https://www.imdb.com/title/{ttId}/ and extract from its static HTML — primarily the <script type="application/ld+json"> block and the <script id="__NEXT_DATA__" type="application/json"> blob. The title page is protected by AWS WAF (AwsWafIntegration token challenge) which returns a 202 with a ~2 KB JS-challenge body to non-browser HTTP clients (including browse cloud fetch, even with --proxies). Drive it from a real browser session: browse open --remote against a Browserbase session created with --verified --proxies. The WAF clears automatically when JS executes.

1. Resolve free-form input → tt-ID (skip if you already have the ID)

The suggestion API is rooted at https://v3.sg.media-imdb.com/suggestion/{firstChar}/{slug}.json. The {firstChar} path component is ignored server-side — any of h, t, or the actual first character of {slug} returns the same response. Build {slug} from the user query by replacing spaces with _ and lowercasing:

SLUG=$(echo "$query" | tr '[:upper:]' '[:lower:]' | tr ' ' '_' | sed 's/[^a-z0-9_]//g')
browse cloud fetch "https://v3.sg.media-imdb.com/suggestion/t/${SLUG}.json"

Response shape — d[] is an ordered list of matches:

{"d":[
  {"id":"tt0111161","l":"The Shawshank Redemption","q":"feature","qid":"movie",
   "rank":78,"s":"Tim Robbins, Morgan Freeman","y":1994,
   "i":{"imageUrl":"https://m.media-amazon.com/...","height":1800,"width":1200}},
  ...
]}

Key fields:

  • id — the tt-prefixed title ID. This is your handoff to step 2.
  • l — title.
  • q — human-readable type ("feature", "TV series", "TV mini-series", "TV episode", "TV short", "short", "TV movie", "video", "podcastSeries", "videoGame").
  • qid — machine type (movie, tvSeries, tvMiniSeries, tvEpisode, tvShort, short, tvMovie, video, podcastSeries, videoGame).
  • y — year (a single integer). Series additionally carry yr as a "YYYY-YYYY" range string (open-ended ongoing series have "YYYY-").
  • rank — IMDb popularity rank (lower = more popular). Do not confuse with the user rating — rank is MOVIEmeter-style popularity, NOT the 0.0-10.0 user score. The user rating is not exposed via the suggestion API at all.
  • s — short top-cast string (comma-separated names, no roles).
  • i — poster image URL + native dimensions.

Disambiguation heuristics (run in order until a single best match is left):

  1. If the input includes a 4-digit year (e.g. "the matrix 1999"), filter d[] to entries where y === year.
  2. If the input includes a type hint ("TV", "series", "movie", "episode", "documentary"), filter d[] by matching qid.
  3. If multiple candidates remain, pick the lowest rank (most popular). If rank is missing on a candidate, treat as Infinity.
  4. If the top two candidates have very close rank values (within 10× of each other) and the query is ambiguous, emit a success: false, reason: "ambiguous_name" result with the top 3-5 candidates rather than guessing.

For TV episodes: the suggestion API surfaces well-known episodes (e.g. "breaking bad ozymandias" → tt2301451) but tends to under-rank lesser-known episode pages. If the query says "season N finale" / "S5E14" / etc. and the suggestion API returns the parent series instead of the episode, fall back to resolving the series first, then navigating to /title/{seriesId}/episodes/?season={N} and reading the episode-list page (or jumping to /title/{episodeId}/).

2. Fetch the title page and extract the rating + metadata

SID=$(browse cloud sessions create --keep-alive --verified --proxies | jq -r '.id')
export BROWSE_SESSION="$SID"
browse open "https://www.imdb.com/title/${ttId}/" --remote
browse wait load --remote
browse wait timeout 1500 --remote     # let lazy hydration settle
HTML=$(browse get html body --remote)

Both --verified and --proxies are required — without them the AWS WAF challenge stalls in browse cloud fetch and IP-blocks/rate-limits a vanilla Browserbase session within a few requests. With them, the title page renders normally (no captcha, no login wall).

2a. Extract <script type="application/ld+json">

The first application/ld+json block on every IMDb title page is a schema.org Movie / TVSeries / TVEpisode object that contains everything you need for the headline rating + most metadata:

{
  "@context": "https://schema.org",
  "@type": "Movie",
  "url": "https://www.imdb.com/title/tt0111161/",
  "name": "The Shawshank Redemption",
  "alternateName": "Cadena perpetua",
  "image": "https://m.media-amazon.com/images/M/MV5B...jpg",
  "datePublished": "1994-10-14",
  "contentRating": "R",
  "duration": "PT2H22M",
  "genre": ["Drama"],
  "aggregateRating": {
    "@type": "AggregateRating",
    "ratingValue": 9.3,
    "ratingCount": 3050000,
    "bestRating": 10,
    "worstRating": 1
  },
  "actor":   [{"@type":"Person","url":"...","name":"Tim Robbins"}, ...],
  "director":[{"@type":"Person","url":"...","name":"Frank Darabont"}],
  "creator": [{"@type":"Organization","url":"..."}, {"@type":"Person","url":"...","name":"Stephen King"}],
  "description": "Over the course of several years, two convicts form a friendship..."
}

Parse it with a hardened regex (NOT JSON.parse on raw HTML; the block may contain HTML-entity-escaped characters in description):

const m = html.match(/<script type="application\/ld\+json">([\s\S]*?)<\/script>/);
const ld = JSON.parse(m[1]);

Field mapping (ld → output JSON):

Output fieldLD-JSON source
titleIdparse from ld.url (/title/(tt\d+)/)
titleld.name
originalTitleld.alternateName if present and !== ld.name, else null
titleTypederive from ld["@type"] (Movie → movie, TVSeries → tvSeries, TVEpisode → tvEpisode, TVMiniSeries → tvMiniSeries, Short → short, VideoGame → videoGame); fall back to __NEXT_DATA__ (see 2b) when @type is generic.
yearyear part of ld.datePublished (or __NEXT_DATA__.releaseYear.year for safety).
yearRangeseries only — from __NEXT_DATA__ (2b).
certificationld.contentRating
runtimeMinutesparse ISO-8601 ld.duration (PT2H22M → 142). Some shorts use PT15M; some series use PT45M as per-episode runtime.
genresld.genre (string → wrap in array)
imdbRatingld.aggregateRating.ratingValue
voteCountld.aggregateRating.ratingCount
actorsld.actor[].name (typically top 5; IMDb truncates here — for the full top-billed list use __NEXT_DATA__, see 2b)
directorsld.director[].name (object or array — normalize to array)
writersld.creator[] filtered to @type === "Person"
posterUrlld.image
shortPlotld.description (HTML-entity-decode after parse)
canonicalUrlld.url

aggregateRating may be absent when a title has fewer than 5 user votes (unrated). Handle missing-gracefully: emit imdbRating: null, voteCount: 0 rather than throwing.

2b. Extract <script id="__NEXT_DATA__" type="application/json">

The LD-JSON block is insufficient for some required fields:

  • Rating distribution per 1-10 bucket (not in LD-JSON at all).
  • Metascore (not in LD-JSON).
  • Full cast list (LD-JSON truncates at ~5).
  • Languages (spokenLanguages).
  • Countries of origin (countriesOfOrigin).
  • TV-episode parent-series ID + season/episode numbers.
  • TV-series year range (endYear).

All of these live in the Next.js page-data blob:

const nm = html.match(/<script id="__NEXT_DATA__" type="application\/json">([\s\S]*?)<\/script>/);
const nd = JSON.parse(nm[1]);
const title = nd.props.pageProps.mainColumnData;   // root for most title fields
const above = nd.props.pageProps.aboveTheFoldData; // root for rating + summary

Useful paths inside mainColumnData / aboveTheFoldData (paths stable across iters; field names match IMDb's internal GraphQL schema):

  • aboveTheFoldData.ratingsSummary.aggregateRating — same ratingValue (decimal).
  • aboveTheFoldData.ratingsSummary.voteCount — same total as LD-JSON.
  • mainColumnData.ratingsSummary.histogram.histogramValues — rating distribution, an array of 10 objects {rating: 10, voteCount: N} from rating 10 down to rating 1. The order is descending — always sort or map by rating rather than relying on positional index.
  • aboveTheFoldData.metacritic.metascore.score — Metascore (or null when no Metascore).
  • mainColumnData.cast.edges[] — full cast; each edge has node.name.nameText.text (actor name), node.characters[].name (role names), node.attributes[].text ("voice", "uncredited", etc.).
  • mainColumnData.principalCredits[] — director/writer/creator grouped by role (category.id === "director" | "writer" | "creator").
  • mainColumnData.spokenLanguages.spokenLanguages[].text — languages.
  • mainColumnData.countriesOfOrigin.countries[].text — countries.
  • mainColumnData.plot.plotText.plainText — short plot (same as LD-JSON description).
  • mainColumnData.outline.plotText.plainText — outline (often null).
  • For series: mainColumnData.releaseYear.year + mainColumnData.releaseYear.endYear (endYear null for ongoing series).
  • For episodes: mainColumnData.series.series.id (parent series tt-ID), mainColumnData.series.series.titleText.text (parent series title), mainColumnData.series.episodeNumber.seasonNumber, mainColumnData.series.episodeNumber.episodeNumber.

For the FULL plot summary (the multi-paragraph "Storyline" block), the __NEXT_DATA__ blob carries it at mainColumnData.summaries.edges[0].node.plotText.plaintext (or null if only a synopsis exists). When you need a longer plot than ld.description, prefer this path.

3. Release the session

browse cloud sessions update "$SID" --body '{"status":"REQUEST_RELEASE"}'

Browser fallback (no API shortcut needed for rating data)

There is no public API surface that returns the IMDb user-rating value. The suggestion API in step 1 is purely a name-resolver. The title-page HTML is the only path to the rating + distribution + Metascore. Don't waste cycles chasing caching.graphql.imdb.com or api.graphql.imdb.com — verified blocked / 500 to anonymous clients (see Site-Specific Gotchas).

Site-Specific Gotchas

  • AWS WAF (AwsWafIntegration) on every www.imdb.com/title/* HTML request from non-browser clients. A bare browse cloud fetch (with or without --proxies) returns HTTP 202 and a ~2 KB body containing an awswaf.com/challenge.js token-acquisition handshake — not the real page. The challenge clears only when JS executes, so the title page must be loaded inside a real browser session. Verified across https://www.imdb.com/title/tt0111161/, /title/.../episodes/, /find/, /_next/data/..., /sitemap.xml, /_json/... — every WAF-protected path returns the same 1991-byte challenge.
  • browse cloud fetch is NOT a viable surface for IMDb title pages. Use it only for the suggestion API (v3.sg.media-imdb.com) and robots.txt — both are WAF-exempt. All www.imdb.com paths the future agent cares about are WAF-protected.
  • Use --verified --proxies on the Browserbase session. Bare sessions get WAF-challenged or IP-rate-limited after a handful of requests. --verified clears the challenge automatically; --proxies rotates the source IP to avoid the rate-limit ban that triggers around request 10-20 from the same datacenter IP.
  • IMDb's robots.txt blocks AI crawlers. Lines User-agent: anthropic-ai / Claude-Web / GPTBot / CCbot / Google-Extended → Disallow: / are present in https://www.imdb.com/robots.txt. The skill must drive a real browser (with a non-bot UA), not curl-fetch with an AI-bot UA. Browserbase's verified-browser path uses a real Chrome UA and clears this.
  • Suggestion-API {firstChar} path component is decorative. https://v3.sg.media-imdb.com/suggestion/h/the_matrix.json and /suggestion/t/the_matrix.json and /suggestion/0/the_matrix.json all return identical JSON. The IMDb search-bar typeahead conventionally sends the first character of the query; the server doesn't care.
  • Suggestion-API rank is MOVIEmeter popularity, NOT user rating. A common trap. The user rating (aggregateRating.ratingValue) is not in the suggestion JSON at all — only the title-page HTML carries it.
  • aggregateRating is missing from the LD-JSON block when a title has fewer than ~5 user votes (typical for obscure shorts, unreleased titles, video-game expansions). Treat as imdbRating: null, voteCount: 0 rather than failing.
  • Rating distribution lives ONLY in __NEXT_DATA__, not in the LD-JSON block. The path is mainColumnData.ratingsSummary.histogram.histogramValues and the array is sorted descending by rating (10 → 1). Always map by rating field; do not assume index 0 == 10.
  • LD-JSON actor array is truncated (typically 5 entries). For the full top-billed cast, parse __NEXT_DATA__.props.pageProps.mainColumnData.cast.edges[].
  • Runtime in LD-JSON is ISO-8601, not minutes. PT2H22M → 142, PT45M → 45. For series, this is the per-episode runtime, not total — note that in the output if the title type is tvSeries/tvMiniSeries.
  • datePublished for series is the series premiere date, not the year range. For a yearRange field on series, read __NEXT_DATA__.mainColumnData.releaseYear.year (start) and .endYear (null for ongoing).
  • Episode pages are also title pages. A TV-episode tt-ID has its own /title/tt.../ page with the same LD-JSON + __NEXT_DATA__ structure. To get parent-series context, read mainColumnData.series.series.id / .titleText.text and mainColumnData.series.episodeNumber.seasonNumber / .episodeNumber.
  • IMDbPro is a different surface (pro.imdb.com). It loads without the WAF challenge but exposes MOVIEmeter / production-contact data, not the public user-rating. Don't use it for rating lookup.
  • IMDb GraphQL is a trap for anonymous clients. Both caching.graphql.imdb.com and api.graphql.imdb.com return 301 → 500 (or block) without a session-cookied request from a logged-in page context. Don't try to bypass the title-page HTML this way.
  • Bulk-data alternative for offline use. IMDb publishes daily TSVs at https://datasets.imdbws.com/ (title.basics.tsv.gz, title.ratings.tsv.gz — only rating + numVotes, no distribution). Useful for batch enrichment of millions of tt-IDs; not appropriate for "what's the rating right now" lookups (24-hour staleness) or for distribution / Metascore (not in the dataset).
  • Read-only — never click Rate, Add to Watchlist, Sign in, or any star-rating bucket. Those mutate user state and require an authenticated user.
  • Original-title detection. ld.alternateName is the original-language title for foreign-language films (e.g. "Cadena perpetua" for tt0111161's Spanish release). It is also populated for some English-language films with regional retitles, so compare alternateName !== name before treating it as "original title".
  • Free-form queries with city/country names don't get rerouted the way OpenTable's term-parser reroutes them — the IMDb suggestion API is purely textual. Safe to pass "Joe's Shanghai" as a movie title without disambiguation tricks.
  • DNS reachability matters for the host environment. The skill requires the executing environment to reach wss://connect.{region}.browserbase.com for CDP traffic, not just api.browserbase.com. Sandboxes that allowlist only the API host can run browse cloud fetch against the suggestion API but cannot drive the title page via browse open --remote — and the title page is mandatory for rating data. Verify CDP reachability before running.

Expected Output

Single, consistent shape — variants by title type are reflected in titleType and the optional seriesContext block.

Movie

{
  "success": true,
  "titleId": "tt0111161",
  "title": "The Shawshank Redemption",
  "originalTitle": null,
  "titleType": "movie",
  "year": 1994,
  "yearRange": null,
  "certification": "R",
  "runtimeMinutes": 142,
  "genres": ["Drama"],
  "imdbRating": 9.3,
  "voteCount": 3050000,
  "ratingDistribution": [
    {"rating": 10, "voteCount": 1830000},
    {"rating": 9,  "voteCount": 580000},
    {"rating": 8,  "voteCount": 320000},
    {"rating": 7,  "voteCount": 150000},
    {"rating": 6,  "voteCount": 70000},
    {"rating": 5,  "voteCount": 38000},
    {"rating": 4,  "voteCount": 18000},
    {"rating": 3,  "voteCount": 12000},
    {"rating": 2,  "voteCount": 8000},
    {"rating": 1,  "voteCount": 25000}
  ],
  "metascore": 82,
  "cast": [
    {"name": "Tim Robbins",     "role": "Andy Dufresne"},
    {"name": "Morgan Freeman",  "role": "Ellis Boyd 'Red' Redding"},
    {"name": "Bob Gunton",      "role": "Warden Norton"},
    {"name": "William Sadler",  "role": "Heywood"},
    {"name": "Clancy Brown",    "role": "Captain Hadley"}
  ],
  "directors": ["Frank Darabont"],
  "writers": ["Stephen King", "Frank Darabont"],
  "posterUrl": "https://m.media-amazon.com/images/M/MV5BMDAyY2FhYjctNDc5OS00MDNlLThiMGUtY2UxYWVkNGY2ZjljXkEyXkFqcGc@._V1_.jpg",
  "shortPlot": "Over the course of several years, two convicts form a friendship, seeking consolation and, eventually, redemption through basic compassion.",
  "fullPlot": "Chronicles the experiences of a formerly successful banker as a prisoner...",
  "languages": ["English"],
  "countries": ["United States"],
  "canonicalUrl": "https://www.imdb.com/title/tt0111161/",
  "seriesContext": null
}

TV Series

{
  "success": true,
  "titleId": "tt11280740",
  "title": "Severance",
  "originalTitle": null,
  "titleType": "tvSeries",
  "year": 2022,
  "yearRange": "2022-",
  "certification": "TV-MA",
  "runtimeMinutes": 60,
  "genres": ["Drama", "Mystery", "Sci-Fi", "Thriller"],
  "imdbRating": 8.7,
  "voteCount": 450000,
  "ratingDistribution": [ {"rating": 10, "voteCount": 0}, ... ],
  "metascore": 87,
  "cast": [ {"name": "Adam Scott", "role": "Mark Scout"}, ... ],
  "directors": [],
  "writers": ["Dan Erickson"],
  "posterUrl": "https://...",
  "shortPlot": "...",
  "fullPlot": "...",
  "languages": ["English"],
  "countries": ["United States"],
  "canonicalUrl": "https://www.imdb.com/title/tt11280740/",
  "seriesContext": null
}

TV Episode

{
  "success": true,
  "titleId": "tt2301451",
  "title": "Ozymandias",
  "originalTitle": null,
  "titleType": "tvEpisode",
  "year": 2013,
  "yearRange": null,
  "certification": "TV-MA",
  "runtimeMinutes": 48,
  "genres": ["Crime", "Drama", "Thriller"],
  "imdbRating": 10.0,
  "voteCount": 250000,
  "ratingDistribution": [ ... ],
  "metascore": null,
  "cast": [ ... ],
  "directors": ["Rian Johnson"],
  "writers": ["Vince Gilligan", "Moira Walley-Beckett"],
  "posterUrl": "https://...",
  "shortPlot": "...",
  "fullPlot": "...",
  "languages": ["English"],
  "countries": ["United States"],
  "canonicalUrl": "https://www.imdb.com/title/tt2301451/",
  "seriesContext": {
    "seriesId": "tt0903747",
    "seriesTitle": "Breaking Bad",
    "seasonNumber": 5,
    "episodeNumber": 14
  }
}

Failure shapes

// Unrated (fewer than ~5 user votes — aggregateRating missing from LD-JSON)
{
  "success": true,
  "titleId": "tt99999999",
  "title": "Some Obscure Short",
  "titleType": "short",
  "imdbRating": null,
  "voteCount": 0,
  "ratingDistribution": [],
  "metascore": null,
  ...
}

// Free-form input could not be confidently resolved to a single tt-ID
{
  "success": false,
  "reason": "ambiguous_name",
  "query": "severance",
  "candidates": [
    {"titleId": "tt11280740", "title": "Severance", "year": 2022, "titleType": "tvSeries", "rank": 150},
    {"titleId": "tt0464196",  "title": "Severance", "year": 2006, "titleType": "movie",    "rank": 8508}
  ]
}

// Free-form input returned zero matches from the suggestion API
{
  "success": false,
  "reason": "title_not_found",
  "query": "ksjdhfksjdhfksjdhf"
}

// WAF challenge could not be cleared (rare with --verified --proxies; document and retry on a fresh session)
{
  "success": false,
  "reason": "anti_bot_block",
  "titleId": "tt0111161",
  "detail": "AWS WAF AwsWafIntegration challenge did not clear after 3 attempts"
}
IMDb Title Rating Lookup · browse.sh