スキル dittobot
✍️

dittobot

中リスク ⚙️ 外部コマンド🌐 ネットワークアクセス📁 ファイルシステムへのアクセス🔑 環境変数⚡ スクリプトを含む

あなたの声を保ちながら文章を書き直す

AIの書き換えツールは、独自性のある声をありきたりで無機質な企業文体にしてしまいがちです。Dittobotは、トーン、事実、リズム、意図を保ちながら、文章をより明確に、シャープに、または自然に仕上げます。

対応: Claude Codex Code(CC)
🥉 75 ブロンズ
1

スキルZIPをダウンロード

2

Claudeでアップロード

設定 → 機能 → スキル → スキルをアップロードへ移動

3

オンにして利用開始

テストする

「dittobot」を使用しています。 これをよりカジュアルに書き直して: 'このたびは応募の状況についてお問い合わせさせていただきたく存じます。15日に書類を提出いたしましたので、ご都合のよろしいときに進捗をご教示いただけますと幸いです。'

期待される結果:

15日に応募書類を提出した件、進捗状況を教えていただけますか? 現状どうなっていますか?

「dittobot」を使用しています。 この文章にパンチを効かせて: 'この製品には、ユーザーが日常業務で便利だと感じる多くの機能があります。'

期待される結果:

この製品が退屈な作業を代わりにやってくれるので、あなたは本業に集中できる。

「dittobot」を使用しています。 この段落の問題点を診断して: '当社は業界のリーディングカンパニーです。多くのソリューションを提供しています。私たちのチームは卓越性に専念しています。多くのセクターにわたる多くのお客様にサービスを提供しています。'

期待される結果:

3つの主な問題: (1) すべての文が同じ出だしで始まっており、平坦なリズムが生まれている。(2) '多くの'、'ソリューション'、'卓越性'のような曖昧な言葉は具体的な意味を持たない。(3) 具体的な証拠、ストーリー、詳細がまったくない。修正方法: 文の冒頭を変える、抽象名詞を具体的な例1つに置き換える、4文のうち少なくとも2文を削除する。

セキュリティ監査

中リスク
v1 • 6/2/2026

Dittobot is a voice-preserving prose rewriting skill with standard installation and evaluation tooling. The static analyzer flagged 490 patterns, but the vast majority are false positives from markdown backtick formatting in documentation files (README, CHANGELOG, CONTRIBUTING, SKILL.md) being misread as shell execution. The 'weak cryptographic algorithm' findings refer to hash operations in Python evaluation/scoring scripts, not security-sensitive cryptography. The install script uses a curl|bash pattern which is conventional for skill installation but worth noting. No confirmed malicious intent; risk is medium due to install scripts, API access in live_eval.py, and overall tooling surface area.

36
スキャンされたファイル
7,423
解析された行数
12
検出結果
1
総監査数

高リスクの問題 (1)

Install script uses curl pipe to shell pattern
README.md:187 documents and install.sh uses a curl|bash installation pattern, which executes remote code directly. While standard for skill distribution, this pattern carries inherent supply chain risk if the remote URL is compromised.
中リスクの問題 (3)
Live evaluation script accesses environment variables for API keys
scripts/live_eval.py reads multiple environment variables that appear to be API keys (for LLM provider authentication). This is expected for a skill that evaluates writing quality via API calls, but it means the skill requires and handles credentials.
Installation scripts perform shell substitution and file operations
install.sh and scripts/install.py use command substitution ($()) and file system operations (shutil, symlink creation, hidden file access in home directory). These are standard for installation tooling but represent execution surface area.
Redaction script references crypto seed patterns for detection
scripts/redact_case.py references crypto seed/private key patterns as part of its redaction logic. This is the tool detecting and redacting sensitive patterns, not storing actual keys, but it shows the script handles sensitive pattern matching.
低リスクの問題 (3)
Hash operations in scoring and evaluation scripts
Multiple Python scripts (scorecard.py, live_eval.py, audit.py, etc.) use hash functions. These are used for content comparison and scoring, not for security-sensitive operations. The static analyzer flags these as 'weak cryptographic algorithm' but they serve non-security purposes.
Markdown backtick formatting detected as shell execution
The overwhelming majority of 'Ruby/shell backtick execution' findings are false positives. Markdown files (README.md, CHANGELOG.md, CONTRIBUTING.md, SKILL.md, etc.) use backticks for inline code formatting, which the static analyzer misinterprets as shell command execution patterns. These are documentation artifacts, not executable code.
Dynamic import expression in report generation script
scripts/rewrite_report.py:14 uses a dynamic import() expression. This appears to be intentional for the skill's report generation module loading, not a security threat.

検出されたパターン

Curl pipe to shell (install pattern)Subprocess execution in build and install scripts
監査者: claude

品質スコア

91
アーキテクチャ
100
保守性
87
コンテンツ
50
コミュニティ
34
セキュリティ
91
仕様準拠

作れるもの

ビジネスメールの推敲

自然な声を保ちながら、仕事のメールをより明確に、自信に満ちた、適切なフォーマルさに書き直す。

ブログ記事やエッセイのブラッシュアップ

ドラフトを引き締め、冗長な部分を削り、パンチを加え、文章がチャットボットではなく本物の人間のように響くようにする。

カバー レターやプロフィールの編集

役職のトーンに合わせながら、個性を残すことで応募書類を際立たせる。

これらのプロンプトを試す

基本的な書き換えリクエスト
このメールをより明確でダイレクトに書き直して。ただし、私の声は保ってね:

[ここにテキストを貼り付け]
トーン調整
このドラフトをより温かく、無機質でないように書き直して。事実と要点はそのまま厳密に保ってください:

[ここにテキストを貼り付け]
詳細な指示付きのフルパンチアップ
このブログ記事セクションを書き直して。目標: 文を短くする、冗長な部分を削る、鮮やかな具体例を一つ加える、私のドライなユーモアを保つ、事実を一切変えない:

[ここにテキストを貼り付け]
診断モード
まだ書き直さないで。この文章で機能していない部分を診断し、テキストから具体的な例を挙げて上位3つの問題をリストアップしてください:

[ここにテキストを貼り付け]

ベストプラクティス

  • 書き直しを依頼する前に、必ずソーステキストまたは確立された声のリファレンスを提供する
  • 望ましい変更の方向性を明確に指定する(より短く、より温かく、より面白く、よりフォーマルに)
  • 変更してはいけないものを明示する: 事実、立場、特定のフレーズ、固有名詞
  • 何を変更すべきか不明な場合は、まず診断モードを使用する

回避

  • ソース素材なしの白紙状態から完全な文章をゼロから作成するようDittobotに依頼しない
  • 事実を捏造したり、元のテキストにない主張を追加するために使用しない
  • 精度が極めて重要で曖昧さが危険を伴う、高度に技術的または法的なテキストには適用しない

よくある質問

Dittobotは標準的なAIリライターと何が違うのですか?
Dittobotは声を置き換えるのではなく、声を守るように設計されています。明瞭さとインパクトを向上させながら、リズム、ユーモア、立場、事実をそのまま保ちます。
ゼロからの文章作成にも使えますか?
Dittobotは、書き直すソーステキストや、合わせるべき確立された声を提供することで最も効果的に機能します。純粋なドラフト作成には、あなたの文章のサンプルやボイスプロファイルを提供してください。
事実や意味を変えてしまいますか?
Dittobotは事実、立場、意図を保つよう指示されています。プロンプトで変更してはいけないものを明示することで、これを強化できます。
どのような種類の文章を扱えますか?
メール、ブログ記事、エッセイ、社内文書、Webサイトのコピー、スピーチ、プロフィール、SNSのキャプション、カバーレター、ソーシャルメディアの投稿など。
複数のスタイル調整を一度に依頼できますか?
はい。1つのプロンプトで「より短く、かつ面白く」や「よりフォーマルだが温かく」のような組み合わせをリクエストできます。
Claude、Codex、Claude Codeで動作しますか?
はい。DittobotはClaude、Codex、Claude Codeと互換性があり、再利用可能なライティングスキルとして機能します。