hangulport

hangulport

hangulport is a client for the hosted Bareun Korean NLP service. It exposes morphological analysis with part-of-speech tagging, the raw morphological model output, Urimalsaem dictionary search by jamo slot pattern, sentence tokenization, spelling and spacing correction, and the service's part-of-speech tag list. One binary serves a command-line interface and a stdio MCP server; both forward every call to the hosted endpoint.

An API key is required. Set BAREUN_API_KEY to your Bareun key; it is sent to the endpoint as the api-key header and nowhere else. Calls go to https://api.bareun.ai/mcp unless BAREUN_MCP_URL overrides the endpoint. When the key is missing, calls fail with an error naming BAREUN_API_KEY — no result is invented locally. The 47 POS tags behind list_pos_tags are served by the upstream; they are not hardcoded here.

Install

npx @reneturcios/hangulport tokenize "안녕하세요"

Or run the container image:

docker run -i ghcr.io/cloudcorprecords/hangulport

The container defaults to the MCP server (see below). Pass CLI arguments to use it as a CLI instead. Both forms need BAREUN_API_KEY in the environment (for Docker, add -e BAREUN_API_KEY=...).

CLI

hangulport analyze_syntax "안녕하세요" --auto-spacing --encoding utf8 --format compact
hangulport analyze_syntax_raw "안녕하세요" --with-sense
hangulport search_dict "ㄱㅏ*" --anchor prefix --limit 5 --with-definition
hangulport tokenize "문장을 나눕니다" --auto-spacing
hangulport correct_grammar "띄어쓰기연습" --enable-sentence-check
hangulport list_pos_tags

hangulport --version

Each subcommand is named after the MCP tool it calls. The first argument is the required text (or dictionary pattern); the remaining flags map one-to-one onto the tool's optional inputs. Boolean flags take no value: passing the flag sets it true, and omitting it forwards nothing — this client injects no defaults of its own. --custom-dict-names takes a comma-separated list. Commands print the upstream result text. Exit codes: 0 on success, 1 when a call fails (including a missing key or an upstream error), 2 for usage errors.

MCP server

One bin serves both faces. The mcp subcommand starts a stdio MCP server:

npx -y @reneturcios/hangulport mcp

MCP client configuration:

{
  "mcpServers": {
    "hangulport": {
      "command": "npx",
      "args": ["-y", "@reneturcios/hangulport", "mcp"],
      "env": { "BAREUN_API_KEY": "your-key" }
    }
  }
}

To point the server at a different Bareun-compatible host, add BAREUN_MCP_URL to the client's env for this server.

Tools

Tool	Input	What it returns
analyze_syntax	text (required); optional auto_split_sentence, auto_spacing, auto_jointing, custom_dict_names, encoding (utf32/utf16/utf8), format (full/compact), with_sense	Morphological analysis with POS tagging.
analyze_syntax_raw	text (required); optional auto_split_sentence, encoding, format, with_sense	Raw morphological model output, no post-processing.
search_dict	pattern (required); optional anchor (word/prefix/suffix/contains), pos, std_only, with_definition, limit, count_only	Urimalsaem dictionary matches for a jamo slot pattern.
tokenize	text (required); optional auto_spacing, encoding	The sentence split into word units.
correct_grammar	text (required); optional custom_dict_names and nine boolean correction options (treat_as_title, disable_split_sentence, disable_caret_spacing, disable_vx_spacing, enable_limited_punctuation, disable_confusion, enable_cleanup_whitespace, disable_typo_correction, enable_sentence_check)	Spelling/spacing correction blocks.
list_pos_tags	(none)	The 47 Bareun POS tags, as served upstream.

Development

npm install
npm run build   # tsc -> dist/
npm test        # node test runner over test/*.test.ts, no network (fake fetch fixtures)
npm run dev     # run the CLI from source via tsx

License

MIT — see LICENSE.