Analyze a document for AI writing artifacts across 7 categories and rewrite it with corrections. Use when a draft needs to be cleaned of detectable AI patterns — overused vocabulary, hedging, structural monotony, voice issues, formatting problems, technical artifacts, and unsupported generalizations. Preserves meaning and author intent throughout.
Check the user's message for a file path.
If no path is provided, use the question tool to ask:
Which file should I refine?
(Allow free-text entry for the file path.)
Read the full file before proceeding.
Work through each category systematically. Count every instance.
Flagged words (replace with context-specific alternatives):
Flagged phrases (remove or rephrase entirely):
[REVIEW: generalization without support][AUTHOR VOICE NEEDED][REVIEW: conclusion restates introduction]Output a table before making any changes:
| Category | Issues found |
|-------------------------|-------------|
| 1. Vocabulary artifacts | N |
| 2. Hedging and filler | N |
| 3. Structural monotony | N |
| 4. Voice and tone | N |
| 5. Heading/formatting | N |
| 6. Technical artifacts | N |
| 7. Deep structural | N |
| **Total** | **N** |
List up to 15 of the most significant changes planned:
Use Edit for targeted fixes or Write for extensive changes. Preserve all meaning, structure, and author intent. Do not add content or expand scope.
Leave markers where author intent is unclear:
[REVIEW: <reason>] — needs human judgment[AUTHOR VOICE NEEDED] — passage requires personal perspective or
domain expertiseDo NOT invent facts, add citations, expand arguments, or change the author's position. The goal is to remove detectable AI artifacts while preserving the document's substance.