Standard Korean Dictionary API for MCP Server

Overview

This MCP (Model Context Protocol) tool provides a Standard Korean Dictionary API that you can host on your MCP server. It lets model agents or local apps query Korean headwords to obtain pronunciations, definitions, grammatical information, and example sentences in a simple JSON format. Integrating this dictionary as an MCP tool makes it easy for LLMs and assistants to enrich responses with authoritative lexical data and usage examples.

The API is designed to be lightweight and RESTful so it can run locally (for privacy) or be deployed to a controlled environment. Registering it as an MCP tool allows language models to call the dictionary during multi-step reasoning, translation, or language-teaching workflows without embedding the entire dictionary into model context.

Features

• Query Korean headwords and receive structured lexical data
• Pronunciation (Hangul + optionally IPA or romanization)
• Multiple senses / definitions per headword
• Part-of-speech and grammatical labels
• Example sentences showing real usage
• Simple JSON REST API suitable for MCP tool integration
• Minimal deployment footprint (Docker-friendly)
• OpenAPI / tool manifest ready for MCP-style registration

Installation / Configuration

Prerequisites:

• Docker (recommended) or a system capable of running the repository service
• MCP server capable of registering external HTTP tools

Clone the repository:

git clone https://github.com/privetin/stdict.git
cd stdict

Run with Docker (example exposes the service on port 8080):

docker build -t stdict .
docker run -p 8080:8080 --rm stdict

If the repository provides a direct startup script (e.g., Node/Python), run it locally:

# Node.js example
npm install
npm start

# or Python example
pip install -r requirements.txt
python app.py

Configuration options (example environment variables):

# sample env variables that the service may support
STDIC_HOST=0.0.0.0
STDIC_PORT=8080
STDIC_LOG_LEVEL=info

Register the service as an MCP tool by adding a tool manifest that points to the service base URL. Example manifest (JSON):

{
  "name": "standard-korean-dictionary",
  "description": "Lookup Korean words: definitions, pronunciation, and examples",
  "base_url": "http://localhost:8080",
  "endpoints": {
    "search": "/api/v1/search?q={query}"
  }
}

Adjust base_url to the public or internal address used by your MCP server.

Available Resources

• GitHub repository: https://github.com/privetin/stdict
• Suggested API endpoints:
  • GET /api/v1/search?q={word}
  • GET /api/v1/entry/{id}
  • (Optional) /openapi.json or /swagger for OpenAPI spec (if provided by the repo)
• Example MCP tool manifest (see above)

API Reference (example)

Search endpoint (HTTP GET)

• Path: /api/v1/search
• Query: q (string) — headword or substring to search
• Returns: JSON array of entries

Example request:

curl "http://localhost:8080/api/v1/search?q=사과"

Example response (trimmed):

[
  {
    "headword": "사과",
    "pronunciation": "사과",
    "pos": ["명사"],
    "senses": [
      {
        "definition": "열매의 일종; 식용",
        "examples": ["사과를 한 개 먹었다.", "사과나무 아래에서 쉬었다."]
      },
      {
        "definition": "잘못에 대한 사죄",
        "examples": ["사과 드립니다."]
      }
    ]
  }
]

Response fields (table)

Use Cases

• Augment chatbots: Let an MCP-enabled assistant fetch precise definitions and examples when asked about Korean vocabulary or usages during a conversation.

  • Example: User asks, “What does 사과 mean?” — the assistant calls the dictionary and returns senses plus example sentences.

• Language learning apps: Provide learners with contextual sentences and POS labels to create flashcards and quizzes.

  • Example flow: Query headword → present definitions and 3 example sentences → generate multiple-choice quiz.

• Translation quality checks: Use dictionary senses to disambiguate potential translations of polysemous words during post-editing or automated checks.

• Text analysis and NLP pipelines: Resolve lemma information and parts of speech for downstream tasks like token normalization, lemmatization, or gloss generation.

• Content generation: Writers or LLMs can ask for natural example sentences to illustrate nuance, register, or collocations.

Quick Integration Tips

• Prefer exact matches for single-headword queries; use substring searches for suggestions/autocomplete.
• Cache frequent requests in your MCP server or agent to reduce latency.
• If integrating into a public-facing MCP setup, add authentication or rate limiting as needed.
• If an OpenAPI spec is present in the repo, use it to auto-generate client bindings or MCP tool manifests.

For full implementation details and any advanced configuration options, consult the repository on GitHub: https://github.com/privetin/stdict.