A specialized AI skill for refining machine-generated academic essays into human-like prose using corpus-informed stylistic weighting.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install essay-humanizer
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install essay-humanizer using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Essay Humanizer?
The Essay Humanizer is a sophisticated tool designed to bridge the gap between detectable AI writing and natural human academic prose. Built as part of the Openclaw Skills ecosystem, it leverages a fine-tuned LoRA adapter over the Qwen3-8B model, specifically optimized for Apple Silicon via the MLX framework. It targets 24 specific linguistic patterns and syntactic markers derived from a statistical analysis of human-written corpora like CAWSE and LOCNESS.
This skill ensures that academic work maintains a professional, human baseline by adjusting vocabulary, punctuation, and rhetorical structures that often trigger AI detection or read as machine-generated. By utilizing data-driven weights, the tool addresses common AI artifacts like cliché metaphors, excessive adverbs, and repetitive sentence structures to produce submission-ready text.
Essay Humanizer Use Cases
- Humanizing AI-drafted argumentative or academic essays to match human stylistic baselines.
- Refining stylistic artifacts in machine-generated content to avoid common AI detection patterns.
- Adjusting syntactic complexity based on Mean Dependency Distance (MDD) targets found in human writing.
- Converting AI-detectable text into professional, continuous prose for academic submission.
- Streamlining the academic writing workflow within Openclaw Skills by automating the polish phase.
How Essay Humanizer Works
- The skill loads a fine-tuned LoRA adapter (9.3 MB) and integrates it with the Qwen3-8B-MLX base model on Apple Silicon.
- The input text is analyzed against a weighted pattern table containing 24 data-driven linguistic markers derived from corpus analysis.
- The inference engine applies syntactic adjustments based on Mean Dependency Distance (MDD) and Average Dependency Distance (ADD) targets to ensure natural sentence variance.
- Higher-priority patterns such as cliché metaphors, em dash overkill, and AI-specific vocabulary are addressed first to maximize the humanizing effect.
- The orchestrator generates a plain-text output that preserves author stance and citations while removing chat-style signatures or instructional artifacts.
Essay Humanizer Setup
To install the Essay Humanizer for Openclaw Skills, run the following commands on an Apple Silicon Mac:
# Install dependencies and decode the adapter
bash scripts/install_deps.sh
# Run the inference script on a draft text file
python scripts/inference.py --file draft.txt
You can also integrate it directly into Python workflows:
from scripts.inference import humanize
print(humanize("Your AI-drafted essay text here..."))
Essay Humanizer Data Schema & Taxonomy
| Component | Path | Description |
|---|---|---|
| Inference script | scripts/inference.py |
The primary entry point for humanization logic. |
| LoRA adapters | assets/adapters/adapters.safetensors.json |
12.3 MB base64 JSON containing the fine-tuned weights. |
| Pattern weights | data/analysis/weights.json |
Statistical weights derived from CAWSE and LOCNESS corpora. |
| Decoder | scripts/decode_adapters.py |
Utility to reconstruct binary safetensors from JSON format. |
| Base model | Qwen/Qwen3-8B-MLX-4bit |
The 4-bit quantized MLX model used as the foundation. |
Essay Humanizer Advanced Features
- Local execution optimized for Apple Silicon using the MLX framework for high-speed inference.
- Data-driven syntactic complexity management using MDD/ADD variance ratios to avoid machine-like sentence smoothing.
- Optional FastAPI host support for remote access or multi-user deployments via HTTP/OpenAPI.
- Automated adapter decoding that reconstructs binary weights on first run for ease of distribution.
- Support for custom pattern weights to prioritize specific stylistic corrections based on different human writing benchmarks.
SKILL.md
Loading
Related Openclaw Skills
.NET Performance Diagnostics & Dump Analyzer
A comprehensive .NET diagnostic toolchain for analyzing CPU hotspots, memory leaks, and GC pressure across Windows, Linux, and macOS.
DOCX Compare / Duplicate Check
An automated tool to identify duplicate text paragraphs and images between two Microsoft Word documents using fuzzy matching and MD5 hashing.
Adult Social Skills
A structured behavioral framework for AI agents to guide users through the logistics, protocols, and maintenance of adult friendships.
DataWorks Daily Task Monitor
An automated monitoring utility that tracks Aliyun DataWorks task statuses and generates detailed daily execution reports with failure alerts.
Postiz CLI
A comprehensive social media automation CLI for scheduling, managing, and analyzing posts across 28+ platforms including X, LinkedIn, and TikTok.
Score
Score is a powerful CLI-based gaming toolkit designed to track results, manage leaderboards, and analyze player statistics through simple text-based logging.
Featured*
