Ultimate Humanizer - Setup Guide¶
Source: surdijon/ultimate-humanizer · 1★ Category: Content & Social / Quality Assurance License: MIT · Published: June 27, 2026
Detects and removes AI-generated writing patterns ("slop") from text. 50 detection patterns, a 2-pass self-audit, and 5-dimensional scoring (naturalness, variety, agency, specificity, rhythm). Optimized for French but works across languages. Includes Google spam-protection mode for SEO content.
Key Capabilities¶
Pattern Detection (50 Patterns)¶
The skill scans for common AI-isms across five categories:
| Pattern Group | Examples Detected | Count |
|---|---|---|
| P1-P14: Rhetorical voids | "In today's digital landscape", "It's worth noting that" | 14 |
| P15-P28: False agency | "I'd be happy to", "Let me walk you through" | 14 |
| P29-P38: Binary contrasts | "On one hand / on the other hand" | 10 |
| P39-P44: Meta-commentary | "To summarize", "In conclusion" | 6 |
| P45-P50: SEO spam markers | Keyword stuffing, thin content signals | 6 |
Intensity Modes¶
| Mode | Patterns | Use Case |
|---|---|---|
--lite |
P1-P14 only | Quick polish, short text |
--full (default) |
P1-P50 | Standard content editing |
--strict |
P1-P50 + active SEO scan | Web/SEO content at risk of Google penalties |
5D Scoring¶
Each text receives scores (0-100) across: - Naturalness - Does it read like a human wrote it? - Variety - Sentence length, structure, vocabulary diversity - Agency - Active voice, concrete subjects, real opinions - Specificity - Facts, numbers, examples vs. vague generalities - Rhythm - Flow, pacing, absence of formulaic transitions
Installation¶
npx skills add surdijon/ultimate-humanizer
Or direct:
git clone https://github.com/surdijon/ultimate-humanizer.git
cp ultimate-humanizer/SKILL.md ~/.hermes/profiles/<your-profile>/skills/ultimate-humanizer.md
Usage with Hermes Agent¶
Activate with the explicit command:
/ultimate-humanizer --full
Or in natural language:
"Humanize this blog post - run the full anti-slop audit"
"Run ultimate-humanizer in strict mode on this landing page"
The skill only activates on explicit invocation - it won't interfere with normal agent operation.
CorpusIQ Use Cases¶
| Use Case | How |
|---|---|
| Blog content QA | Run --strict on all published posts before going live |
| Email copy editing | --lite pass removes AI-isms from templated emails |
| SEO content audit | --strict mode catches patterns that trigger Google spam filters |
| Client deliverables | Ensure human-quality tone in all external-facing content |
This guide is part of the Hermes Skills Catalog. Discovered June 29, 2026. Powered by CorpusIQ.
Integration with CorpusIQ¶
CorpusIQ is the connector layer that lets AI assistants read the systems a business already runs on - accounting, payments, CRM, email, files and analytics - and answer with the source named. It is read-only by design: it retrieves data and never writes to, or acts in, a connected system, which is what makes it safe to point at live accounts.
This page is part of the CorpusIQ directory, the reference we maintain for the MCP server and agent-skill ecosystem. Run alongside CorpusIQ, a third-party server like this one can be combined with a company's own business data in a single answer.