← Back to Skills Marketplace
artur-zhdan

Humanize AI

by artur-zhdan · GitHub ↗ · v1.1.0
cross-platform ✓ Security Clean
2316
Downloads
8
Stars
3
Active Installs
2
Versions
Install in OpenClaw
/install humanize-ai
Description
Humanize AI content by detecting and auto-fixing AI generated content. Humanize AI text using Python scripts. Scans for AI vocabulary, puffery, chatbot artifacts, and auto-replaces filler phrases. Use when you want to analyze text in AI detector and bypass it in future, batch-process files, run automated cleanup, or get a report before manual humanizing.
README (SKILL.md)

Humanize CLI

Command-line tools for detecting and auto-fixing AI writing patterns.

Scripts

analyze.py — Detect AI Patterns

Scans text and reports AI vocabulary, puffery, chatbot artifacts, and auto-replaceable phrases.

# Analyze a file
python scripts/analyze.py input.txt

# Analyze from stdin
echo "This serves as a testament to our commitment" | python scripts/analyze.py

# JSON output for programmatic use
python scripts/analyze.py input.txt --json

Output example:

==================================================
AI PATTERN ANALYSIS - 5 issues found
==================================================

AI VOCABULARY:
  • testament: 1x
  • crucial: 2x

AUTO-REPLACEABLE:
  • "serves as" → "is": 1x
  • "in order to" → "to": 1x

humanize.py — Auto-Replace Patterns

Performs automatic replacements for common AI-isms.

# Humanize and print to stdout
python scripts/humanize.py input.txt

# Write to output file
python scripts/humanize.py input.txt -o output.txt

# Include em dash replacement
python scripts/humanize.py input.txt --fix-dashes

# Quiet mode (no change log)
python scripts/humanize.py input.txt -q

What it fixes automatically:

  • Filler phrases: "in order to" → "to", "due to the fact that" → "because"
  • Copula avoidance: "serves as" → "is", "boasts" → "has"
  • Sentence starters: removes "Additionally,", "Furthermore,", "Moreover,"
  • Curly quotes → straight quotes
  • Chatbot artifacts: removes "I hope this helps", "Let me know if", etc.

Workflow

  1. Analyze first to see what needs fixing:

    python scripts/analyze.py document.txt
    
  2. Auto-fix safe replacements:

    python scripts/humanize.py document.txt -o document_clean.txt
    
  3. Manual review for AI vocabulary and puffery flagged by analyze (these require human judgment)

  4. Re-analyze to confirm improvements:

    python scripts/analyze.py document_clean.txt
    

Customizing Patterns

Edit scripts/patterns.json to add/remove:

  • ai_words — vocabulary that flags but doesn't auto-replace
  • puffery — promotional language to flag
  • replacements — phrase → replacement mappings (empty string = delete)
  • chatbot_artifacts — phrases to auto-remove
  • hedging_phrases — excessive hedging to flag

Batch Processing

Process multiple files:

# Analyze all markdown files
for f in *.md; do
  echo "=== $f ===" 
  python scripts/analyze.py "$f"
done

# Humanize all txt files in place
for f in *.txt; do
  python scripts/humanize.py "$f" -o "$f.tmp" && mv "$f.tmp" "$f"
done

Usage Guidance
This skill appears to do what it says: local detection and replacement of AI-style wording. Before installing or running it: 1) Test the scripts on small sample files and inspect changes (use -o to write to a separate file) rather than running in-place batch loops; 2) Backup important files before using the included batch examples that overwrite originals (the provided for-loop example uses mv to replace files); 3) Review and customize scripts/patterns.json to ensure the replacement rules match your intent (some replacements delete text and some patterns are broad); 4) Note minor implementation bugs (the curly-quotes handling and some quote-counting code appear incorrect/ineffective) — consider reviewing/fixing those functions if you rely on them; 5) Be aware the description explicitly mentions 'bypass' AI detectors: the tool provides text-level modifications that can change detector output (dual-use) — consider ethical and policy implications for your use case.
Capability Analysis
Type: OpenClaw Skill Name: humanize-ai Version: 1.1.0 The skill bundle is designed to analyze and humanize text content using Python scripts. While it declares the `Shell` tool in `SKILL.md`, which is a high-risk capability, its usage is transparently demonstrated for executing the Python scripts (`analyze.py`, `humanize.py`) and performing basic file iteration, which is directly aligned with the stated purpose of processing local files. The Python scripts themselves only perform local text processing, reading from specified input files or stdin, and writing to specified output files or stdout, without any evidence of data exfiltration, malicious execution, persistence mechanisms, or deceptive prompt injection.
Capability Assessment
Purpose & Capability
Name/description claim: detect and auto-fix AI wording. Provided artifacts (patterns.json, analyze.py, humanize.py) implement detection and auto-replacement locally. No unrelated credentials, binaries, or external services are requested.
Instruction Scope
SKILL.md instructs running local Python scripts and batch shell loops that may overwrite files in place (the example uses a temp file then mv back). That behavior is coherent with the purpose but can cause data loss if run without backups. The description mentions 'bypass' AI detectors which is a dual-use/ethical note but the code itself only does local textual replacements and reporting (no detector API calls or exfiltration).
Install Mechanism
Instruction-only skill with bundled Python scripts; no install spec, no external downloads, and no package installation. Low risk from installation mechanism.
Credentials
No environment variables, credentials, or config paths are requested. The scripts only read/write files provided by the user — access is proportional to the stated functionality.
Persistence & Privilege
always:false and agent-autonomy defaults are unchanged. The skill does not request persistent presence or modify other skill/system configs.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install humanize-ai
  3. After installation, invoke the skill by name or use /humanize-ai
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.1.0
- Skill renamed from "humanize-cli" to "humanize-ai" - Description updated to reflect the new name and clarify purpose - No code or functional changes detected
v1.0.0
Initial release of pythonic humanize-cli with detection and auto-fix tools: - Added analyze.py: Detects AI vocabulary, puffery, artifacts, and auto-replaceable phrases in text files or stdin. - Added humanize.py: Auto-replaces AI-isms (filler phrases, bot artifacts, wordy starters, etc.) and outputs humanized text. - Added patterns.json: Centralized pattern list for detection and replacements, customizable by user. - Supports plain and JSON output, quiet/logging options, and batch processing via shell. - Enables workflow: analyze → auto-fix → manual review → re-analyze.
Metadata
Slug humanize-ai
Version 1.1.0
License
All-time Installs 4
Active Installs 3
Total Versions 2
Frequently Asked Questions

What is Humanize AI?

Humanize AI content by detecting and auto-fixing AI generated content. Humanize AI text using Python scripts. Scans for AI vocabulary, puffery, chatbot artifacts, and auto-replaces filler phrases. Use when you want to analyze text in AI detector and bypass it in future, batch-process files, run automated cleanup, or get a report before manual humanizing. It is an AI Agent Skill for Claude Code / OpenClaw, with 2316 downloads so far.

How do I install Humanize AI?

Run "/install humanize-ai" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Humanize AI free?

Yes, Humanize AI is completely free (open-source). You can download, install and use it at no cost.

Which platforms does Humanize AI support?

Humanize AI is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Humanize AI?

It is built and maintained by artur-zhdan (@artur-zhdan); the current version is v1.1.0.

💬 Comments