Browse ghibli films
When the user asks about Studio Ghibli films, movies, directors, Miyazaki, Takahata, Rotten Tomatoes scores, release years, or wants to browse the Ghibli catalog — reach for the Studio Ghibli API. 22 films, unauthenticated, no search endpoint.
browse-ghibli-films
· v1
· updated 2026-04-16
When to use this skill
When the user asks about Studio Ghibli films — titles, directors, Rotten Tomatoes scores, release years, running times — or wants to browse or sort the catalog. The dataset is only 22 films but richer than it looks: Grave of the Fireflies is a Takahata-directed WWII drama with a higher rt_score than most Miyazaki entries. For linked character, species, location, or vehicle records, follow the URL arrays in each film. For box office figures or streaming availability, this is the wrong skill.
Your best first call
curl "https://ghibliapi.vercel.app/films"
No auth. No key. Returns all 22 films in one unpaginated response. Fetch this once per conversation, index by id, and filter client-side — there is no search endpoint; all narrowing is local.
Key fields per film:
id — UUID, used in /films/{id} and referenced by cross-reference endpointstitle — English display titleoriginal_title — Japanese title in kanji/kanaoriginal_title_romanised — romanized Japanese titledirector — not always Miyazaki; Isao Takahata directed several entries including the highest-rated film in the catalogrelease_date, running_time, rt_score — all strings; parse to integer before sortingpeople, species, locations, vehicles — arrays of resource URLs; follow each to get the linked record
Fallbacks (when the best call isn't enough)
- A specific film by UUID →
/films/{id} returns a single film record. Use when you already have the id from a prior catalog fetch.
- Cross-referenced characters, species, locations, or vehicles → Follow the URLs in the
people, species, locations, or vehicles arrays. Each resolves to a full record with its own cross-links back to the film.
Pitfalls
- Every numeric-looking field —
rt_score, running_time, release_date — is a string. "9" > "97" lexicographically. Parse to integer before any comparison or sort.
- A
people or locations element of the form https://ghibliapi.vercel.app/people/ (trailing slash, no UUID) means the film has no linked records in that category. Following it returns the full people list, not an empty result — silently wrong data.
- The dataset is frozen at 22 films. Earwig and the Witch (2021) is the most recent entry; The Boy and the Heron (2023) is absent.
One-line summary for the user
I can browse the Studio Ghibli film catalog — titles, directors, RT scores, release years, running times — via a single unauthenticated call, but all numeric fields come back as strings and the dataset is frozen at 22 films through 2021.
SKILL.md source (frontmatter + body)
---
name: browse-ghibli-films
description: When the user asks about Studio Ghibli films, movies, directors, Miyazaki, Takahata, Rotten Tomatoes scores, release years, or wants to browse the Ghibli catalog — reach for the Studio Ghibli API. 22 films, unauthenticated, no search endpoint.
---
## When to use this skill
When the user asks about Studio Ghibli films — titles, directors, Rotten Tomatoes scores, release years, running times — or wants to browse or sort the catalog. The dataset is only 22 films but richer than it looks: Grave of the Fireflies is a Takahata-directed WWII drama with a higher `rt_score` than most Miyazaki entries. For linked character, species, location, or vehicle records, follow the URL arrays in each film. For box office figures or streaming availability, this is the wrong skill.
## Your best first call
```bash
curl "https://ghibliapi.vercel.app/films"
```
No auth. No key. Returns all 22 films in one unpaginated response. Fetch this once per conversation, index by `id`, and filter client-side — there is no search endpoint; all narrowing is local.
Key fields per film:
- `id` — UUID, used in `/films/{id}` and referenced by cross-reference endpoints
- `title` — English display title
- `original_title` — Japanese title in kanji/kana
- `original_title_romanised` — romanized Japanese title
- `director` — not always Miyazaki; Isao Takahata directed several entries including the highest-rated film in the catalog
- `release_date`, `running_time`, `rt_score` — all strings; parse to integer before sorting
- `people`, `species`, `locations`, `vehicles` — arrays of resource URLs; follow each to get the linked record
## Fallbacks (when the best call isn't enough)
- **A specific film by UUID** → `/films/{id}` returns a single film record. Use when you already have the `id` from a prior catalog fetch.
- **Cross-referenced characters, species, locations, or vehicles** → Follow the URLs in the `people`, `species`, `locations`, or `vehicles` arrays. Each resolves to a full record with its own cross-links back to the film.
## Pitfalls
- Every numeric-looking field — `rt_score`, `running_time`, `release_date` — is a string. `"9" > "97"` lexicographically. Parse to integer before any comparison or sort.
- A `people` or `locations` element of the form `https://ghibliapi.vercel.app/people/` (trailing slash, no UUID) means the film has no linked records in that category. Following it returns the full people list, not an empty result — silently wrong data.
- The dataset is frozen at 22 films. Earwig and the Witch (2021) is the most recent entry; The Boy and the Heron (2023) is absent.
## One-line summary for the user
I can browse the Studio Ghibli film catalog — titles, directors, RT scores, release years, running times — via a single unauthenticated call, but all numeric fields come back as strings and the dataset is frozen at 22 films through 2021.