/ ディレクトリ / プレイグラウンド / tutor-skills
● コミュニティ bevibing ⚡ 即起動

tutor-skills

作者 bevibing · bevibing/tutor-skills

Two skills — tutor-setup turns PDFs/docs/codebases into an Obsidian study vault; tutor then quizzes you with progress tracking and spaced review.

Pairs content ingestion with interactive quizzing. tutor-setup scans your sources (PDFs, text, HTML, or code), extracts concepts with verified source mapping, and builds an Obsidian vault with wiki-linked notes, diagrams, and questions. tutor runs rounds of 4 questions with emoji-tracked mastery, diagnostic sessions, drills, and hard-mode reviews.

なぜ使うのか

主な機能

ライブデモ

実際の動作

tutor-skill.replay ▶ 準備完了
0/0

インストール

クライアントを選択

~/Library/Application Support/Claude/claude_desktop_config.json  · Windows: %APPDATA%\Claude\claude_desktop_config.json
{
  "mcpServers": {
    "tutor-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/bevibing/tutor-skills",
        "~/.claude/skills/tutor-skills"
      ],
      "_inferred": true
    }
  }
}

Claude Desktop → Settings → Developer → Edit Config を開く。保存後、アプリを再起動。

~/.cursor/mcp.json · .cursor/mcp.json
{
  "mcpServers": {
    "tutor-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/bevibing/tutor-skills",
        "~/.claude/skills/tutor-skills"
      ],
      "_inferred": true
    }
  }
}

Cursor は Claude Desktop と同じ mcpServers スキーマを使用。プロジェクト設定はグローバルより優先。

VS Code → Cline → MCP Servers → Edit
{
  "mcpServers": {
    "tutor-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/bevibing/tutor-skills",
        "~/.claude/skills/tutor-skills"
      ],
      "_inferred": true
    }
  }
}

Cline サイドバーの MCP Servers アイコンをクリックし、"Edit Configuration" を選択。

~/.codeium/windsurf/mcp_config.json
{
  "mcpServers": {
    "tutor-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/bevibing/tutor-skills",
        "~/.claude/skills/tutor-skills"
      ],
      "_inferred": true
    }
  }
}

Claude Desktop と同じ形式。Windsurf を再起動して反映。

~/.continue/config.json
{
  "mcpServers": [
    {
      "name": "tutor-skill",
      "command": "git",
      "args": [
        "clone",
        "https://github.com/bevibing/tutor-skills",
        "~/.claude/skills/tutor-skills"
      ]
    }
  ]
}

Continue はマップではなくサーバーオブジェクトの配列を使用。

~/.config/zed/settings.json
{
  "context_servers": {
    "tutor-skill": {
      "command": {
        "path": "git",
        "args": [
          "clone",
          "https://github.com/bevibing/tutor-skills",
          "~/.claude/skills/tutor-skills"
        ]
      }
    }
  }
}

context_servers に追加。保存時に Zed がホットリロード。

claude mcp add tutor-skill -- git clone https://github.com/bevibing/tutor-skills ~/.claude/skills/tutor-skills

ワンライナー。claude mcp list で確認、claude mcp remove で削除。

ユースケース

実用的な使い方: tutor-skills

Turn a 400-page textbook PDF into a study vault you can actually use

👤 Students and self-learners ⏱ ~90 min intermediate

使うタイミング: You have a textbook and would rather quiz yourself than re-read it.

前提条件
  • Obsidian installed — obsidian.md — free
  • Claude Code CLI — Required for running the skills
フロー
  1. Run tutor-setup
    Use tutor-setup on ./textbook.pdf — build the vault at ~/vaults/calc.✓ コピーしました
    → Vault created with chapter notes, concept maps, and question bank
  2. Start a study session
    Use tutor. Start a diagnostic session on chapter 3.✓ コピーしました
    → 4 questions with grading and mastery update
  3. Review weak areas
    Show my weak concepts and drill them.✓ コピーしました
    → Targeted drill on lowest-mastery concepts

結果: A personal study vault plus a working retention loop.

注意点
  • PDFs with poor OCR yield low-quality notes — Run OCR with a better tool first if the source is image-based
組み合わせ: human-skill-tree-skill

Onboard to an unfamiliar codebase via guided quizzing

👤 New hires and contractors ⏱ ~120 min intermediate

使うタイミング: You inherited a 50k-line codebase and reading it linearly isn't working.

フロー
  1. Ingest the repo
    Use tutor-setup on ~/src/acme-api. Trace the architecture and build onboarding exercises.✓ コピーしました
    → Vault with module map, architecture diagrams, and exercise questions
  2. Run onboarding exercises
    Use tutor. Start the onboarding track.✓ コピーしました
    → Exercises that force you to navigate the code to answer

結果: Faster ramp-up than reading docs cover-to-cover.

注意点
  • Generated exercises too shallow — Ask tutor-setup to focus on specific modules rather than the whole repo
組み合わせ: filesystem

組み合わせ

他のMCPと組み合わせて10倍の力を

Document-centric (tutor) + dialogue-centric (human-skill-tree)

Use tutor on my textbook, then switch to human-skill-tree for Socratic discussion of the hard concepts.✓ コピーしました
tutor-skill + filesystem

Save progress summaries outside the vault for version control

Every week, snapshot my mastery report to ~/learning-log/.✓ コピーしました

ツール

このMCPが提供する機能

ツール入力呼び出すタイミングコスト
tutor-setup source files (PDF/text/HTML/code) Start of a new study topic LLM tokens for extraction
tutor (quiz rounds) vault + focus topic Regular study sessions Tokens per round
Diagnostic mode topic Start of topic or after time away Tokens
Drill mode weak concepts Mid-study reinforcement Tokens
Hard-mode review mastered concepts Confidence check before exam Tokens

コストと制限

運用コスト

APIクォータ
None
呼び出しあたりのトークン
Setup can be heavy (20-100k tokens for a big PDF). Quizzes are light (2-5k per round).
金額
Free — skill is local. Obsidian is free.
ヒント
Split large textbooks by chapter for setup to keep ingestion manageable.

セキュリティ

権限、シークレット、影響範囲

認証情報の保管: No credentials — skill is prompts + local file ops
データ送信先: None — everything runs locally

トラブルシューティング

よくあるエラーと対処法

Vault notes missing concepts from the source

Re-run tutor-setup with chapter-level focus; full-book in one pass can skim.

Questions feel too easy

Switch to hard-mode review or ask tutor for application-level questions.

Obsidian doesn't show the wiki-links

Check vault settings; enable backlinks and graph view plugins.

代替案

tutor-skills 他との比較

代替案代わりに使う場面トレードオフ
human-skill-tree-skillYou prefer dialogue-based learning without a document vaultDifferent modality — no persistent vault
AnkiYou want pure SRS without the AI tutoring layerStronger SRS, no content generation

その他

リソース

📖 GitHub の公式 README を読む

🐙 オープンな issue を見る

🔍 400以上のMCPサーバーとSkillsを見る