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

rootnode-skills

作者 drayline · drayline/rootnode-skills

19 skills for diagnosing, building, and optimizing Claude Projects — prompt compilation, anti-pattern detection, context budget analysis.

drayline/rootnode-skills is a suite for engineering quality Claude Projects. Build skills (prompt-compilation, skill-builder), Diagnose skills (project-audit, global-audit, anti-pattern-detection, prompt-validation), Optimize skills (behavioral-tuning, memory-optimization, context-budget), plus 4 block libraries (identity, reasoning, output formats) and 5 domain packs (strategy, SWE, content, research, agents).

なぜ使うのか

主な機能

ライブデモ

実際の動作

rootnode-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": {
    "rootnode-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/drayline/rootnode-skills",
        "~/.claude/skills/rootnode-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": {
    "rootnode-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/drayline/rootnode-skills",
        "~/.claude/skills/rootnode-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": {
    "rootnode-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/drayline/rootnode-skills",
        "~/.claude/skills/rootnode-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": {
    "rootnode-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/drayline/rootnode-skills",
        "~/.claude/skills/rootnode-skills"
      ],
      "_inferred": true
    }
  }
}

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

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

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

claude mcp add rootnode-skill -- git clone https://github.com/drayline/rootnode-skills ~/.claude/skills/rootnode-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: rootnode-skills

Diagnose why a Claude Project stopped following instructions

👤 Claude.ai power users whose Project drifted ⏱ ~45 min intermediate

使うタイミング: Your Project worked, you added stuff, now it ignores key rules.

前提条件
  • Skill cloned — git clone https://github.com/drayline/rootnode-skills ~/.claude/skills/rootnode-skills
フロー
  1. Run the project audit
    Use rootnode-project-audit on my Project system prompt + instructions.✓ コピーしました
    → Six-dimension scorecard with specific failures called out
  2. Check for anti-patterns
    Now rootnode-anti-pattern-detection — which of the 7 patterns do I have?✓ コピーしました
    → Named patterns (e.g. 'conflicting directives', 'buried critical instruction')
  3. Rebalance memory
    Apply rootnode-memory-optimization — where should each rule actually live?✓ コピーしました
    → Recommendation on what belongs in system prompt vs project docs vs user memory

結果: A diagnosed Project with a concrete fix list.

注意点
  • Audit report is long; you only fix the easy ones — Start with anti-patterns flagged — those block the rest

Compile a new Claude Project system prompt from scratch

👤 Anyone building a Claude Project for the first time ⏱ ~60 min intermediate

使うタイミング: You have goals, constraints, and examples — but not yet a prompt.

フロー
  1. Run prompt-compilation
    rootnode-prompt-compilation — here's my intent, constraints, and 3 examples. Parse / Select / Construct / Validate.✓ コピーしました
    → Structured system prompt following the four stages
  2. Validate
    Now rootnode-prompt-validation — score it across the six dimensions.✓ コピーしました
    → Scorecard with specific weaknesses
  3. Apply behavioral tuning
    rootnode-behavioral-tuning — which of my instructions will Claude drift from?✓ コピーしました
    → Countermeasures inserted at the right places

結果: A well-engineered Project prompt on day one.

組み合わせ

他のMCPと組み合わせて10倍の力を

rootnode-skill + skillcheck-free-skill

Audit your Project prompt, then validate any skills it references against spec

Run rootnode-project-audit, then SkillCheck on each referenced SKILL.md.✓ コピーしました
rootnode-skill + planning-with-files

Audit produces a long fix list — track progress on it across sessions

Use planning-with-files to track the rootnode audit fix list across this week.✓ 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
rootnode-prompt-compilation intent + constraints + examples Starting a new Project or major rewrite 0
rootnode-project-audit existing Project prompt Project behavior degraded 0
rootnode-anti-pattern-detection prompt/instructions When Claude ignores rules 0
rootnode-memory-optimization all prompt + project files Information is in the wrong place 0
rootnode-context-budget project files list Hitting context limits 0
rootnode-behavioral-tuning instructions You need reliable adherence 0

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
Audits read whole prompts — can be large
金額
Free — skills are local files
ヒント
Run audit once per major revision, not per-message.

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 — pure prompts
データ送信先: None

トラブルシューティング

よくあるエラーと対処法

Skill not invoked — Claude gives generic prompt advice instead of using the audit

Reference the specific sub-skill by name: 'rootnode-project-audit'.

確認: ls ~/.claude/skills/rootnode-skills/
SKILL.md frontmatter wrong in a sub-skill

Each sub-skill has its own SKILL.md — inspect the folder.

確認: head ~/.claude/skills/rootnode-skills/rootnode-project-audit/SKILL.md
Wrong trigger keywords — wrong sub-skill activates

Sub-skill names are specific ('project-audit' vs 'global-audit' vs 'full-stack-audit'); use the right one.

代替案

rootnode-skills 他との比較

代替案代わりに使う場面トレードオフ
skills-skill-8 (huangwb8)You want skill-authoring tooling rather than Project-level engineeringDifferent layer — skill vs Project
Raw prompt iterationSmall Projects that don't need formal engineeringLower overhead, higher drift risk

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