Bareun MCP server

What it does

This MCP server exposes the Bareun Korean NLP engine to MCP-compatible AI tools, providing morphological analysis, tokenization, spell and grammar correction, and dictionary search for Korean text. It targets developers and AI agent builders who need accurate Korean language processing beyond what general-purpose LLMs provide.

• Correct spacing, typos, and grammar in Korean text and receive structured correction blocks explaining each change.
• Analyze Korean sentences into morphemes with part-of-speech tags, optionally including homograph sense disambiguation.
• Search the Urimalsaem Korean dictionary (~1.1M entries) using phoneme-level jamo patterns to find words by ending, coda, or other sub-syllable conditions.

Tools this server exposes

Tool	Does
analyze_syntax	Splits Korean text into morphemes and tags parts of speech, with optional preprocessing, custom dictionaries, and homograph sense disambiguation.
analyze_syntax_raw	Performs morphological analysis without any post-processing, returning raw model output without compound splitting, auto-spacing, or custom dictionaries.
search_dict	Searches the Urimalsaem Korean dictionary using jamo phoneme-level slot patterns with filters for part of speech, standard forms, and definitions.
tokenize	Splits a Korean sentence into word-level tokens.
correct_grammar	Corrects Korean spelling, spacing, and grammar and returns structured correction blocks with the original text, corrected text, category, and rule for each fix.
list_pos_tags	Returns the 47 part-of-speech tag codes, names, and classes used by Bareun.

Before you enable it

An MCP server runs with whatever access you give it. Start with a development account rather than production, and check which tools the server exposes before you let an agent call them unattended.