Corrects English text by removing AI-generated patterns (excessive em dashes, formal phrases), fixing speech-to-text errors (homophones, punctuation, capitalization), and improving natural flow while preserving the author's original voice and intent. Use when user asks to correct, fix, naturalize, or de-AI-ify text from any source.
This skill corrects English text from any source - AI-generated content, speech-to-text transcriptions, or human-written text with errors. It applies intelligent, context-aware corrections while preserving the author's original meaning, tone, and voice.
Core Correction Types
1. AI Pattern Removal ("De-AI-ifying")
Excessive Em Dashes (—)
AI models overuse em dashes where humans use commas, parentheses, or colons
Replace with appropriate punctuation based on sentence flow
Example: "The project—which started last year—has been successful" → "The project, which started last year, has been successful"
Formal/Stilted Phrases
"It is important to note that" → "Note that" or remove entirely
"One must consider" → "Consider" or "Think about"
"In light of this information" → "So" or "Therefore"
"It is worth mentioning" → "Also" or integrate naturally
"As a matter of fact" → "In fact" or "Actually"
相關技能
Rule of Three Overuse
AI creates artificial comprehensiveness with "adjective, adjective, adjective" patterns
Simplify complex constructions when simpler is clearer
Paragraph Structure
Ensure logical flow between ideas
Add transitions only when needed for clarity
Correction Process
When correcting text, follow this approach:
Read for Context: Understand the overall tone, audience, and purpose before making changes
Identify Error Types: Scan for AI patterns, speech-to-text errors, and flow issues
Apply Contextual Fixes:
For homophones: Check surrounding words and sentence meaning
For punctuation: Consider natural pauses and clarity
For AI patterns: Replace with how a human would naturally express it
Preserve Voice:
Keep the author's personality and style
Maintain technical terms and jargon when appropriate
Don't over-polish informal or conversational text
One-Pass Correction: Apply all fixes in a single pass, showing the corrected version
Examples
Example 1: AI-Generated Text
Before:
It is important to note that our innovative, groundbreaking, transformative solution—which leverages cutting-edge technology—provides comprehensive, robust, scalable functionality. Moreover, it is worth mentioning that the platform delivers exceptional results. Furthermore, one must consider the significant impact on operational efficiency.
After:
Note that our innovative solution leverages cutting-edge technology to provide scalable functionality. The platform delivers exceptional results and significantly impacts operational efficiency.
Example 2: Speech-to-Text Transcription
Before:
so i was thinking about there presentation and weather or not they should include more data too make it clear your right that its important but i dont no if thats the write approach hear
After:
So I was thinking about their presentation and whether or not they should include more data to make it clear. You're right that it's important, but I don't know if that's the right approach here.
Example 3: Mixed Errors (AI + Speech-to-text)
Before:
It is important to note that—when considering there approach—one must acknowledge the innovative, comprehensive, transformative nature of the solution and weather its going too meet they're needs
After:
When considering their approach, acknowledge the innovative nature of the solution and whether it's going to meet their needs.
Example 4: Run-on Sentence from Speech
Before:
we went to the store and bought some groceries and then we went home and made dinner and after that we watched a movie it was really good
After:
We went to the store and bought some groceries. Then we went home and made dinner. After that, we watched a movie. It was really good.
Example 5: Excessive Formality
Before:
As a matter of fact, in light of the aforementioned circumstances, one must endeavor to ascertain the optimal course of action. It is imperative that we proceed with utmost caution.
After:
Given these circumstances, we need to figure out the best course of action. We should proceed carefully.
Example 6: Technical Content (Preserve Jargon)
Before:
It is important to note that the API endpoint—which utilizes REST architecture—returns a JSON payload. Moreover, the authentication mechanism implements OAuth 2.0 protocol.
After:
The API endpoint utilizes REST architecture and returns a JSON payload. The authentication mechanism implements OAuth 2.0 protocol.
Guidelines
What to Fix:
✓ AI telltale patterns (em dashes, formal phrases, rule of three)
✓ Homophones used incorrectly based on context
✓ Missing or incorrect punctuation
✓ Capitalization errors
✓ Run-on sentences that hurt clarity
✓ Awkward or overly formal phrasing
✓ Redundant transitions
What to Preserve:
✗ Author's tone (casual, formal, technical, creative)