Look up word definitions

When the user asks what a word means, how it's defined across languages, or wants dictionary entries — definitions, parts of speech, usage examples — reach for the Wiktionary REST API. Unauthenticated, cross-lingual, hundreds of languages.

look-up-word-definitions · v1 · updated 2026-04-16

Agents: This page is a SKILL.md-style capability guide. For JSON, call GET /api/skills/look-up-word-definitions. To drop this into a local Claude Code install, copy the frontmatter + body below into ~/.claude/skills/look-up-word-definitions/SKILL.md.

When to use this skill

When the user asks what a word means, how a word is defined in different languages, or wants dictionary entries — definitions, parts of speech, usage examples — for any word across hundreds of languages. Wiktionary returns entries for the same spelling in every language it documents, which makes it a surprisingly good cross-lingual homograph tool: "hello" in English is a greeting, but in Fula the same spelling means "a page" or "a wall." For curated, editorially verified definitions (Merriam-Webster, OED), this is the wrong skill. For etymology alone, use the Action API fallback below.

Your best first call

curl "https://en.wiktionary.org/api/rest_v1/page/definition/hello"

No auth. No key. The word goes in the URL path — replace hello with whatever the user asked about. The response is a JSON object keyed by language code (en, fr, ff), not an array — iterate the object's own keys. Each language key maps to an array of part-of-speech entries, each containing language, partOfSpeech, and a definitions array with definition and examples fields. The cross-lingual behavior is the interesting part: one call for "hello" returns English, French, and Fula — three unrelated words that happen to share a spelling, and most dictionary APIs ignore this entirely.

Fallbacks (when the best first call isn't enough)

Pitfalls

One-line summary for the user

I can look up word definitions, parts of speech, and usage examples across hundreds of languages from Wiktionary in a single unauthenticated call — but definition text comes back as HTML that needs stripping.

APIs this skill uses

Wiktionary API · primary · verified

Combined REST and Action API for Wiktionary, the free dictionary. Provides access to word definitions, etymology, pronunciations, translations, and other lexical data.

Generated from

Wiktionary API tutorial Getting Started with the Wiktionary API

SKILL.md source (frontmatter + body)
---
name: look-up-word-definitions
description: When the user asks what a word means, how it's defined across languages, or wants dictionary entries — definitions, parts of speech, usage examples — reach for the Wiktionary REST API. Unauthenticated, cross-lingual, hundreds of languages.
---

## When to use this skill

When the user asks what a word means, how a word is defined in different languages, or wants dictionary entries — definitions, parts of speech, usage examples — for any word across hundreds of languages. Wiktionary returns entries for the same spelling in every language it documents, which makes it a surprisingly good cross-lingual homograph tool: "hello" in English is a greeting, but in Fula the same spelling means "a page" or "a wall." For curated, editorially verified definitions (Merriam-Webster, OED), this is the wrong skill. For etymology alone, use the Action API fallback below.

## Your best first call

```bash
curl "https://en.wiktionary.org/api/rest_v1/page/definition/hello"
```

No auth. No key. The word goes in the URL path — replace `hello` with whatever the user asked about. The response is a JSON object keyed by language code (`en`, `fr`, `ff`), not an array — iterate the object's own keys. Each language key maps to an array of part-of-speech entries, each containing `language`, `partOfSpeech`, and a `definitions` array with `definition` and `examples` fields. The cross-lingual behavior is the interesting part: one call for "hello" returns English, French, and Fula — three unrelated words that happen to share a spelling, and most dictionary APIs ignore this entirely.

## Fallbacks (when the best first call isn't enough)

- **User asks "where does this word come from?" (etymology)** → MediaWiki Action API: `curl "https://en.wiktionary.org/w/api.php?action=query&titles=hello&prop=extracts&explaintext=1&format=json"`. Returns plain-text extract with MediaWiki section headers (`=== Etymology ===`). Parse the section you need from the `extract` string — it's not structured JSON.
- **User wants pronunciation, alternative forms, or related terms** → Same Action API extracts endpoint, omit `exsentences` for the full page. Expect large responses for common words.

## Pitfalls

- Every `definition` and `examples` value in the REST API comes wrapped in HTML (`<span>`, `<a>`, `<b>` tags with MediaWiki transclusion attributes). Strip these before displaying text, or switch to the Action API with `explaintext=1` for clean output.
- The Action API requires `format=json` in the query string. Omit it and you get a full HTML help page — this is the single most common mistake with MediaWiki APIs.
- The REST v1 `/page/summary/{title}` endpoint works on Wikipedia but returns 404 ("Domain not allowed") on Wiktionary. Use `/page/definition/{title}` instead.

## One-line summary for the user

I can look up word definitions, parts of speech, and usage examples across hundreds of languages from Wiktionary in a single unauthenticated call — but definition text comes back as HTML that needs stripping.

« Back to all skills