Polyglottika Polyglottika
HomeGet startedPricingBusinessAPIMCPBlogArtistsCareersContactStart Learning
Start Learning

Model Context Protocol

Coming soon

Give your AI the keys to the Rift.

Tools, resources, and prompts so an assistant can work directly on your subtitle data and language databases — without copy-paste.

01 · Functions

MCP Tools (Actionable Functions for AI Models)

Tools allow an AI assistant (like Claude Desktop or a custom MCP client) to execute functions directly on your subtitle data or language databases:

  1. Actionable function parse_subtitles

    Subtitle & Script Ingestion Tool

    What it does

    Accepts raw .srt, .vtt, or dialogue script text, strips out timestamp/formatting clutter, and chunks the content into clean, sentence-level dialogue blocks ready for extraction.

  2. Actionable function extract_vocabulary

    Vocabulary & Lemma Tool

    What it does

    Analyzes a submitted subtitle line or paragraph, extracts unknown words, determines their dictionary base (lemma) forms, and fetches definitions, readings (Furigana/Pinyin/Romanization), and frequency data.

  3. Actionable function analyze_grammar

    Syntax & Pattern Engine

    What it does

    Mimics a deep syntax breakdown tool. It takes a complex sentence (e.g., Japanese, Korean, Mandarin, French) and parses out particles, verb conjugations, tenses, and honorific levels, explaining why a sentence is structured that way.

  4. Actionable function create_scene_card

    Flashcard Generator

    What it does

    Automatically formats a captured word, its English/native translation, and the exact subtitle/scene context into a structured flashcard object ready for Spaced Repetition Systems (SRS) or database sync.

  5. Actionable function transfer_kanji_hanja

    Cross-Language Etymology Tool

    What it does

    Links shared characters or roots across languages (e.g., mapping Japanese Kanji to Korean Hanja and Chinese Hanzi) to leverage cross-language transfer when a user studies multiple languages on the platform.

02 · Context

MCP Resources (Read-Only Data Contexts)

Resources allow the AI model to directly read and reference your personal learning data or shared media repositories without manually copying and pasting:

Read-only context polyglottika://user/vocabulary

User vocabulary graph

Exposes the user's active/learned word list, flashcard decks, and retention stats so the AI can tailor its explanations to the user's actual proficiency level.

Read-only context polyglottika://media/scene-library

Scene library

Provides access to community-vetted subtitle scenes, game scripts, or story chapters for contextual querying (e.g., "Find me 3 examples of this grammar pattern used in our imported anime scripts").

03 · Workflows

MCP Prompts (Pre-built Workflows)

Prompts act as reusable templates that users can quickly trigger inside an MCP-compatible chat interface:

Pre-built workflow immersion_drill

Immersion drill

Generates a quick conversational roleplay or shadow-reading drill using the exact vocabulary and grammar lines from the user's latest subtitle import session.

Pre-built workflow ceo_explanation

CEO explanation

Formats grammar and syntax feedback using a specific persona template (matching Polyglottika's "Language CEO" styling—such as strict syntax logic, classical elegance, or mythic storytelling).