Skills detect-ai
๐Ÿ”

detect-ai

Low Risk ๐ŸŒ Network access๐Ÿ”‘ Env variables

Detect AI-Generated Text

This skill analyzes text to detect if it was written by AI, returning a score from 0-100 with detailed metrics. Use it to verify content authenticity before publishing or submitting.

Supports: Claude Codex Code(CC)
โš ๏ธ 67 Poor
1

Download the skill ZIP

2

Upload in Claude

Go to Settings โ†’ Capabilities โ†’ Skills โ†’ Upload skill

3

Toggle on and start using

Test it

Using "detect-ai". /detect-ai The quick brown fox jumps over the lazy dog. This is a simple sentence that demonstrates basic English grammar and syntax structure.

Expected outcome:

## AI Detection Results

**Score:** 12/100 (Human-written)
**Words Analyzed:** 23

### Metrics
- Perplexity: 18
- Burstiness: 25
- Readability: 45
- N-gram Score: 12

### Recommendation
This content appears to be human-written. No humanization needed.

Using "detect-ai". /detect-ai In today's rapidly evolving technological landscape, the integration of artificial intelligence systems has become increasingly prevalent across diverse sectors. This transformative approach enables organizations to optimize operations while simultaneously reducing operational costs through automated processes.

Expected outcome:

## AI Detection Results

**Score:** 87/100 (Highly likely AI-generated)
**Words Analyzed:** 42

### Metrics
- Perplexity: 94
- Burstiness: 12
- Readability: 28
- N-gram Score: 6

### Recommendation
This content shows strong AI-generated patterns. Consider using HumanizerAI to humanize before publishing.

Security Audit

Low Risk
v1 โ€ข 2/27/2026

Static analysis flagged multiple patterns but evaluation reveals false positives. The detected 'external_commands' are markdown code blocks, not shell execution. The 'network' and 'env_access' findings are legitimate API calling patterns - the skill makes expected HTTP requests to humanizerai.com with an API key for authentication. No actual security risks identified.

1
Files scanned
90
Lines analyzed
4
findings
1
Total audits
Low Risk Issues (2)
External API Access
Skill makes HTTP requests to external API endpoint https://humanizerai.com/api/v1/detect. This is expected behavior for an API-based skill.
Environment Variable Access for API Key
Skill accesses HUMANIZERAI_API_KEY from environment for authentication. This is standard practice for API-based skills.

Risk Factors

๐ŸŒ Network access (2)
๐Ÿ”‘ Env variables (2)

Detected Patterns

False Positive: Shell Command DetectionFalse Positive: Weak Cryptographic AlgorithmFalse Positive: Obfuscation Heuristic
Audited by: claude

Quality Score

38
Architecture
100
Maintainability
87
Content
23
Community
86
Security
91
Spec Compliance

What You Can Build

Content Verification for Publishers

Publishers use this skill to verify submitted articles are original and not AI-generated, ensuring content quality standards.

Academic Integrity Checking

Educators and institutions use this skill to check student submissions for AI-generated work.

Pre-Submission Quality Check

Writers use this skill to check their own content before submitting to ensure it does not appear AI-generated.

Try These Prompts

Basic AI Detection
/detect-ai [paste your text here]
Analysis with Context
Use the detect-ai skill to analyze the following text and explain what the different metrics mean: [text content]
Batch Content Check
Analyze this articleๆฎต่ฝ for AI-generated content: [paste article]. Give me a summary of the verdict and key metrics.
Recommendation Request
/detect-ai [text]. Based on the score, should I use the HumanizerAI tool to humanize this content before publishing?

Best Practices

  • Use the score.overall value (not 'score') when presenting results to users
  • Provide context by explaining what each metric means (perplexity, burstiness, readability)
  • Always include actionable recommendations based on the score range

Avoid

  • Displaying the raw JSON response directly to users without interpretation
  • Ignoring API errors and not providing helpful troubleshooting
  • Treating the score as definitive proof rather than a probability indicator

Frequently Asked Questions

What does the AI detection score mean?
The score ranges from 0-100. Lower scores (0-40) indicate human-written content, middle scores (41-60) are uncertain, and higher scores (61-100) indicate AI-generated content.
What API key do I need?
You need a HUMANIZERAI_API_KEY from https://humanizerai.com. Set it in your environment variables before using the skill.
Can this skill guarantee if content is AI-written?
No. The skill provides a probability estimate based on linguistic patterns. It is not definitive proof of AI generation.
What are the key metrics shown?
The main metrics are perplexity (language model uncertainty), burstiness (variation in sentence length), readability (text complexity), and n-gram score (pattern analysis).
What if the API call fails?
Check that HUMANIZERAI_API_KEY is set correctly. Verify the API service is available. The skill provides error messages for debugging.
Can this detect edited AI content?
Detection accuracy decreases for heavily edited AI content. The skill may show lower scores for humanized text.

Developer Details

File structure

๐Ÿ“„ SKILL.md