Letterboxd movie lookup
This pack gives an agent movie metadata, ratings, directors, genres, synopsis, and user reviews from Letterboxd. It uses a built-in fetch() and regex scraper, with optional delegation to a recipe-capable scraping tool.
Installation
npm install @framers/agentos-ext-letterboxd
Peer dependency: @framers/agentos with range >=0.10.40.
Usage
import { AgentOS } from '@framers/agentos';
const agentos = await AgentOS.create({
extensionManifest: {
packs: [{ package: '@framers/agentos-ext-letterboxd' }],
},
});
Configuration
| Option | Secret id | Environment variable | Required | Default |
|---|---|---|---|---|
recipeTool | None | None | No | Built-in scraper |
priority | None | None | No | 50 |
The optional recipeTool exposes execute({ recipe, params }, context). The tool calls it with the letterboxd-movie recipe, a query parameter, and a string year when supplied. A failed recipe or missing output data falls back to the built-in scraper.
Tools
| Tool | What it does |
|---|---|
letterboxd_movie | Looks up a movie by title and optional year, returning film metadata and a sample of user reviews. |
letterboxd_movie
| Parameter | Type | Required | Description |
|---|---|---|---|
title | string | Yes | Film title to search for on Letterboxd. |
year | integer | No | Optional release year to disambiguate titles with the same name. |
Notes
The built-in scraper needs no API key or headless browser. It searches film pages, prefers a result whose surrounding HTML mentions the supplied year, and otherwise selects the first result. When search yields no slug, it tries a film URL derived from the title.
An empty title returns a tool error. Unavailable pages can return found: false; successful pages can contain only the fields the parser finds. The tool has no write side effects and declares capability:media_search. The built-in scraper uses MAX_REVIEWS = 5 and truncates each review's text with slice(0, 500). These limits do not apply to reviews returned by the recipe tool.
License
Apache-2.0