| name | utility-authenticity-audit |
| description | Use when user types /utility-authenticity-audit or asks for this novel workflow command. Audit a text for AI-generated stylistic patterns and cliches. |
/utility-authenticity-audit
Treat text after /utility-authenticity-audit as $ARGUMENTS.
User Input: $ARGUMENTS
Objective
Analyze text for AI-generated stylistic patterns, flag unnatural phrases, and suggest concrete fixes.
Execution Steps
1. Identify Target
Determine what text the user wants to audit:
- Full chapter — e.g., "Chapter 3" → read
./stories/[novel-name]/content/chapter_[N].md
- Line range — e.g., "Chapter 3, lines 15–30" → read that section
- Pasted text — use the text provided directly in the command input
If the target is unclear, ask the user to specify.
2. Analyze for AI Patterns
Scan the text and flag:
- Clichés: overused phrases like "a testament to", "shivers down the spine", "a stark reminder", "tapestry of"
- Structural patterns: overly symmetrical paragraphs, repetitive sentence rhythms, formulaic paragraph openings
- Emotional descriptors: naming emotions directly instead of showing through physical sensation or behavior
- Generic language: vague descriptions that could apply to any scene ("the bustling market", "the ancient forest")
- AI rhythm: sentences that feel too evenly paced, lacking the natural irregularity of human writing
3. Generate Report
Provide a structured report:
- Flagged Items: list each problematic phrase/pattern with its location (line number or paragraph)
- Why It Feels Unnatural: explain why each flagged item reads as AI-generated
- Suggested Fixes: provide 3 concrete ways to inject grit, specificity, and human authenticity into each flagged section
4. Output
Present the audit report. Do NOT auto-rewrite — present findings and wait for user direction.
