Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.openhome.com/llms.txt

Use this file to discover all available pages before exploring further.

Movie Recommender is an OpenHome community ability for discovering movies by voice. It helps users find recommendations, trending titles, similar movies, release dates, ratings, summaries, and streaming options using The Movie Database (TMDB).

What It Does

  • Recommends movies by genre, mood, year, or free-text request
  • Finds trending and top-rated movies
  • Finds similar movies from a title the user likes
  • Gives details about a specific movie from the current results or by title
  • Answers release-date questions, including upcoming movie phrasing
  • Speaks ratings with a natural quality label like excellent, solid, mixed, or rough
  • Looks up US streaming providers through TMDB watch providers
  • Lets the user ask for more results without starting over
  • Summarizes the current movie picks
  • Handles vague requests with a clarifying question instead of a weak search
  • Resolves ordinals and pronouns like the second one, it, and that one

Supported Requests

Request typeExampleWhat happens
RecommendationRecommend something scarySearches TMDB for matching movies and speaks a few picks
TrendingWhat's trending in movies?Returns currently trending movies
Top-ratedBest movies of all timeReturns highly rated movies
Similar moviesMovies like InceptionFinds the source title, then recommends similar titles
DetailsTell me about the second oneOpens a specific movie and speaks a short synopsis
Release dateWhen does it come out?Gives the release date for a named or focused movie
RatingWhat's that rated?Speaks the TMDB score with a short quality description
Watch providersWhere can I watch it?Lists available US streaming providers from TMDB
More resultsShow me morePages through the current search
SummariesWhat are these about?Gives short overviews of current picks

Example Prompts

  • “What’s trending in movies?”
  • “Find me some best movies.”
  • “Recommend something scary.”
  • “Sci-fi releases in 2025.”
  • “Movies like The Matrix.”
  • “Tell me about the second one.”
  • “When does it come out?”
  • “What’s that rated?”
  • “Where can I watch it?”

Example Conversation

User: Recommend something scary. AI: Got it, finding a few that fit. Here are a few that fit: The Substance from 2024, rated 7.5. Smile 2 from 2024, rated 7.0. Longlegs from 2024, rated 6.6. User: Tell me about the second one. AI: Smile 2, from 2024, rated 7.0. A pop star battles an entity that twists her perception of reality before her world tour. User: Where can I watch it? AI: Smile 2 is on Paramount Plus.

Trigger Phrases

  • “movie recommender”
  • “recommend a movie”
  • “what should I watch”
  • “trending movies”
  • “movie suggestion”

Data Source

SourceOpenHome API key nameRole
The Movie Database (TMDB)tmdb_api_keyMovie discovery, metadata, ratings, release dates, similar titles, watch providers

Setup

Add your TMDB v3 API key in OpenHome Settings → API Keys as tmdb_api_key. Use the short TMDB v3 API key only. Do not use the API Read Access Token or API Secret, and do not hardcode the key.

Getting a TMDB API Key

  1. Create or sign in to a TMDB account.
  2. Open your account settings.
  3. Go to the API section.
  4. Request API access if it is not already enabled.
  5. Choose Developer access and fill out the required application details.
  6. After approval, copy the short key labeled API Key (v3 auth).
  7. In OpenHome, open Settings -> API Keys.
  8. Add a new key named tmdb_api_key.
  9. Paste the TMDB v3 API key as the value and save it.
If TMDB rejects the key, confirm that the saved value is the short v3 API key and that there are no extra spaces.

Voice UX Notes

  • Short fillers before TMDB calls so the user knows the ability is working.
  • Pre-formatted spoken lines — no LLM naturalize pass between filler and result, which keeps mic-hot time short.
  • Continue prompts taper — turn 1 is a rich invitation, turn 2 is a shorter directional prompt, turn 3 onward listens silently.
  • Exit routing owned by the LLM — “stop”, “okay stop”, “I’m done” are all recognized.

Developer Credit

Developed by @Kaushal-205.

View on GitHub

Source code for the movie-recommender community ability.