analyze-dependencies
Analyze project dependencies and evaluate architectural health. Trigger with "analyze dependencies", "detect circular dependencies", "architecture issues?",…
Detect mechanical AI patterns and improve to natural text. Trigger with "check for AI-ness", "detect mechanical writing", "make it sound natural", "fix AI-like text".
$ npx -y skills add wasabeef/claude-code-cookbook --skill style-ai-writing --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/style-ai-writingContext preview
The summary Claude sees to decide when to auto-load this skill.
Detect mechanical AI patterns and improve to natural text. Trigger with "check for AI-ness", "detect mechanical writing", "make it sound natural", "fix AI-like text".
description: 'Detect mechanical AI patterns and improve to natural text. Trigger with "check for AI-ness", "detect mechanical writing", "make it sound natural", "fix AI-like text".' allowed-tools: - Read
Detects mechanical patterns in AI-generated text and provides suggestions for improving to more natural Japanese.
/ai-writing-check [options]
# Check AI writing style in file cat README.md /ai-writing-check "Check this document for AI writing style and suggest improvements" # Analyze specific file /ai-writing-check --file docs/guide.md "Detect AI-like expressions and suggest corrections to natural expressions" # Scan entire project /ai-writing-check --dir . --severity high "Report only critical AI writing issues in the project"
Examples detected: - **Important**: This is an important item - Completed item (with checkmark emoji) - Hot topic (with fire emoji) - Ready to start (with rocket emoji) Improved examples: - Important item: This is an important item - Completed item - Notable topic - Ready to start
Examples detected: Revolutionary technology will change the industry. This completely solves the problem. Works like magic. Improved examples: Effective technology brings change to the industry. Solves many problems. Works smoothly.
Examples detected: **Idea**: New proposal (with lightbulb emoji) **Caution**: Important warning (with warning emoji) Improved examples: Idea: New proposal Note: Important warning
Examples detected: First, we will perform the setup. You can use this tool. Performance is greatly improved. Improved examples: First, perform setup. You can use this tool. Performance improves by 30%.
# Analyze entire document for AI writing style cat article.md /ai-writing-check "Analyze and suggest improvements from these perspectives: 1. Detection of mechanical expressions 2. Suggestions for correction to natural Japanese 3. Priority-based improvement list" # Focus on specific patterns /ai-writing-check --file blog.md "Pay special attention to exaggerated and redundant expressions and suggest improvements" # Batch check multiple files find . -name "*.md" -type f /ai-writing-check --dir docs/ "Analyze AI writing style throughout the documentation and create a summary"
# Compare before and after improvement /ai-writing-check --file draft.md "Detect AI-like expressions and present them in the following format: - Problem areas (with line numbers) - Type of problem and reason - Specific improvement suggestions - Effect of improvement" # Auto-fix mode /ai-writing-check --file report.md --fix "Automatically fix detected patterns and report results" # Project AI writing style report /ai-writing-check --dir . --severity all "Analyze AI writing style throughout the project and provide: 1. Statistical information (detection count by pattern) 2. Top 5 most problematic files 3. Improvement priority matrix 4. Step-by-step improvement plan"
# Apply custom rules /ai-writing-check --file spec.md "Check technical specifications with these additional criteria: - Ambiguous expressions (appropriate, as needed) - Lack of specificity (fast → specific numbers) - Inconsistent terminology usage" # Check for CI/CD integration /ai-writing-check --dir docs/ --severity high "Output results in GitHub Actions executable format: - Number of errors and filenames - Line numbers requiring correction - Exit code configuration" # Style guide compliance check /ai-writing-check --file manual.md "Additional checks based on company style guide: - Honorific usage (unification of desu/masu form) - Appropriate use of technical terms - Consideration for readers"
When you run the `/ai-writing-check` command, Claude performs the following processes:
1. **Pattern Detection**: Detects AI-like patterns from specified files or text 2. **Specific Correction Suggestions**: Presents correction suggestions with line numbers for each issue 3. **--fix Mode**: Automatically fixes detected patterns and displays a summary of results 4. **Report Generation**: Provides detection count, improvement priority, and comparison before/after correction
Claude reads the actual file contents and performs analysis based on the rules of textlint-rule-preset-ai-writing.
This command is created with reference to the [textlint-rule-preset-ai-writing](https://github.com/textlint-ja/textlint-rule-preset-ai-writing) rule set. It is a textlint rule preset for detecting mechanical patterns in AI-generated text and promoting more natural expressions.
A collection of commands, roles, and automation scripts for Claude Code. Automate your workflow without unnecessary confirmations, allowing you to focus on what matters.
Repo: wasabeef/claude-code-cookbook
Analyze project dependencies and evaluate architectural health. Trigger with "analyze dependencies", "detect circular dependencies", "architecture issues?",…
Performance analysis based on Core Web Vitals with UX scoring. Trigger with "analyze performance", "improve speed", "check Core Web Vitals", "page speed",…
Verify information accuracy against codebase and documentation. Trigger with "is this correct?", "fact check", "verify this", "is this accurate?".
Evaluate and improve AI prompt quality. Trigger with "check this prompt", "evaluate prompt quality", "improve this prompt".
Generate commit messages from staged changes. Trigger with "suggest commit message", "generate commit message", "what should the commit say?", "write commit…