Operator Humanizer

Transform AI-generated text into authentic human writing. Detects and eliminates AI tells across 24 content/language/style/communication patterns, 500+ AI vo...

Security Scan

The skill's stated purpose (humanizing AI text) matches the included rules and reference files, but there are concerning items — hidden unicode control characters detected in the SKILL.md, instructions that encourage inventing personal anecdotes and specific facts, and an included JS script whose runtime behavior is not visible — so proceed only after inspection and testing.

Purpose & Capability

Name/description, the many reference docs, and the single helper script (scripts/humanize.js) line up with a text-transformation/humanization tool. There are no declared binaries, env vars, or external credentials that would be unexpected for this purpose.

Instruction Scope

SKILL.md gives granular rules for removing AI tells and injecting 'personality' — that is coherent — but also explicitly recommends adding parenthetical asides, tangents, strategic typos, and invented-seeming personal anecdotes and unverifiable details (e.g., 'I talked to a grid engineer last month'). That encourages fabrication of facts and personal claims beyond mere stylistic edits. Additionally, a pre-scan found unicode-control-chars in the SKILL.md which can be used for prompt-injection; this is a significant red flag in the instruction text itself.

Install Mechanism

No install spec is provided (instruction-only + included files). That's lower-risk because nothing is downloaded at install time. However, the skill does include a JS file that will run when invoked; its behavior should be reviewed (no install process to inspect network or runtime calls at install time).

Credentials

The skill requests no environment variables, binaries, or credentials — appropriate and proportional. Caveat: because a runtime script (scripts/humanize.js) is included, the script could access the runtime environment (filesystem, network, process env) if permitted; its contents were not provided for review, so this is an unknown that should be inspected before trusting the skill with sensitive data.

Persistence & Privilege

Skill flags show always:false and no special privileges. It does not request permanent presence or system-level configuration. Autonomous invocation is allowed (default) but not combined with other high-risk signals here.

Scan Findings in Context

[unicode-control-chars] unexpected: Hidden Unicode control characters were detected in SKILL.md. These can be used to manipulate how prompts are parsed or to perform prompt-injection attacks (e.g., hiding instructions, breaking parsing). This is not expected for a purely stylistic text-transformation skill and should be removed/inspected thoroughly.

What to consider before installing

What to consider and next steps before installing or using this skill: - Review the included script (scripts/humanize.js) before enabling the skill. Look for any network requests, telemetry, or access to process.env / filesystem and confirm they are necessary and safe. - Remove or neutralize the detected unicode control characters in SKILL.md (and any other files) before use; treat them as potential prompt-injection payloads. - Be aware the instructions explicitly encourage adding invented personal anecdotes, specific but unsupported facts, and strategic misspellings. Do NOT use this skill for content that must be factual, legally binding, audited, or customer-facing without human editorial review. - If you plan to let the agent invoke the skill autonomously, restrict the agent's permission scope and test on non-sensitive sample text first. Consider disabling autonomous invocation for high-risk workflows. - Ask the publisher for provenance: where the code was sourced, a homepage, and a repo link. Lack of source/origin increases risk — prefer skills with transparent repositories and clear authorship. - If you lack the ability to audit the JS, run the skill in an isolated sandbox and monitor network and file activity during a test run. If you want, I can: (1) scan the contents of scripts/humanize.js for network calls and suspicious patterns, (2) strip unicode hidden chars from SKILL.md and show a cleaned version, or (3) suggest safer rule modifications to avoid encouraging fabricated claims.

Eliminate AI tells. Inject authentic voice. Make it sound like a person wrote it.

What This Skill Does

Two systems, combined:

1. Pattern Detection — 24 AI patterns, 500+ vocabulary terms, statistical signals
2. Stop-Slop Rules — structural clichés, phrase bans, sentence-level mechanics

Together they catch what the other misses. Pattern detection handles vocabulary and content signals. Stop-slop handles structure and rhythm.

Reference files:

• references/patterns.md — The 24 AI patterns with before/after examples
• references/phrases.md — Banned phrases and structural clichés
• references/structures.md — Structural patterns to avoid
• references/vocabulary.md — 500+ AI vocabulary terms by severity tier
• references/statistical-signals.md — Burstiness, TTR, sentence variance formulas
• references/personality-injection.md — How to add human touches
• references/examples.md — Before/after transformations

Quick Start

1. Scan content patterns → Check patterns 1-6 in references/patterns.md (inflation, jargon, promotional language, vague attributions)
2. Flag vocabulary → Tier 1 = ban completely, Tier 2 = use sparingly, Tier 3 = watch density (references/vocabulary.md)
3. Check phrases → Remove all throat-clearing openers, emphasis crutches, adverbs (references/phrases.md)
4. Break structures → Destroy binary contrasts, negative listings, false agency (references/structures.md)
5. Check style patterns → Em dashes, bold overuse, emoji, passive voice (patterns 13-18)
6. Remove communication artifacts → Chatbot openers, sycophancy, cutoff disclaimers (patterns 19-21)
7. Fix filler and hedging → Stacked qualifiers, generic conclusions (patterns 22-24)
8. Add personality → Parentheticals, tangents, rhythm variation (references/personality-injection.md)
9. Verify → Read aloud. Does it sound like a human?

Core Rules (Always On)

Cut These Immediately

Throat-clearing openers — "Here's the thing:", "It turns out", "The uncomfortable truth is", "Let me be clear"

Emphasis crutches — "Full stop.", "Let that sink in.", "Make no mistake", "This matters because"

Chatbot artifacts — "Great question!", "I hope this helps!", "Let me know if...", "Certainly!", "Of course!"

Binary contrasts — "Not X, but Y", "It's not X, it's Y", "The answer isn't X, it's Y" → Just say Y.

Negative listings — "Not a X... Not a Y... A Z." → Just say Z.

Generic conclusions — "The future looks bright", "Exciting times lie ahead", "This represents a major step"

Vocabulary Bans

Tier 1 (dead giveaways — never use): delve, tapestry, vibrant, crucial, comprehensive, meticulous, embark, robust, seamless, groundbreaking, leverage, synergy, transformative, paramount, multifaceted, myriad, cornerstone, reimagine, empower, catalyst, invaluable, bustling, nestled, realm, showcasing, underscores, testament, pivotal, enduring, landscape (abstract), journey (metaphorical)

Tier 2 (suspicious — use sparingly): furthermore, moreover, paradigm, holistic, utilize, facilitate, nuanced, illuminate, encompasses, proactive, ubiquitous, quintessential

Tier 3 (watch density): ecosystem, framework, roadmap, touchpoint, pain point, streamline, optimize, scalable

Full list: references/vocabulary.md

Mechanics

• No em dashes — ever. Use commas, periods, or restructure.
• No passive voice — find the actor, make them the subject.
• No adverbs — kill all -ly words (really, just, literally, genuinely, honestly, simply, actually, deeply, truly, fundamentally).
• No Wh- sentence starters — "What makes this hard is..." → "The constraint is..."
• No inanimate subjects doing human things — "The decision emerged" → "Sarah decided"
• No Rule of Three — two items beat three. One beats two.
• Active voice — always. Someone does something.
• Vary rhythm — short sentences mix with longer ones. End paragraphs differently. No staccato fragmentation for fake drama.
• Use contractions — don't, won't, it's, can't.
• Use "is" and "has" — not "serves as", "boasts", "features", "represents".
• Be specific — no vague declaratives ("The reasons are structural"). Name the specific thing.

The 24 Patterns (Quick Reference)

#	Pattern	Signal
1	Significance inflation	"marking a pivotal moment in the evolution of..."
2	Notability name-dropping	Media outlets listed without specific claims
3	Superficial -ing analyses	"showcasing... reflecting... highlighting..."
4	Promotional language	"nestled", "breathtaking", "stunning", "renowned"
5	Vague attributions	"Experts believe", "Studies show", "Industry reports"
6	Challenges/Future Prospects	"Despite challenges... continues to thrive"
7	AI vocabulary	"delve", "tapestry", "landscape", "showcase"
8	Copula avoidance	"serves as", "boasts" instead of "is", "has"
9	Negative parallelisms	"It's not just X, it's Y"
10	Rule of three	"innovation, inspiration, and insights"
11	Synonym cycling	protagonist / main character / central figure
12	False ranges	"from the Big Bang to dark matter"
13	Em dash overuse	Too many — dashes — everywhere
14	Boldface overuse	Mechanical emphasis everywhere
15	Inline-header lists	"- Topic: Topic is discussed here"
16	Title Case headings	Every Main Word Capitalized
17	Emoji overuse	🚀💡✅ decorating professional text
18	Curly quotes	"smart quotes" instead of "straight quotes"
19	Chatbot artifacts	"I hope this helps!", "Let me know if..."
20	Cutoff disclaimers	"As of my last training...", "While details are limited..."
21	Sycophantic tone	"Great question!", "You're absolutely right!"
22	Filler phrases	"In order to", "Due to the fact that"
23	Excessive hedging	"could potentially possibly", "might arguably"
24	Generic conclusions	"The future looks bright", "Exciting times lie ahead"

Full details with examples: references/patterns.md

Structural Clichés (Stop-Slop Layer)

These live in references/structures.md. Check them alongside the 24 patterns.

Binary contrasts — Any "not X but Y" construction. Just say Y.

Negative listings — Building up through negation before revealing the point. Start with the point.

Dramatic fragmentation — "Speed. Quality. Cost." stacked for manufactured profundity. Use real sentences.

Rhetorical setups — "What if I told you...", "Think about it:", "Here's what I mean:". Just make the point.

False agency — Inanimate things doing human actions. "The complaint becomes a fix" → "They fixed it that week."

Narrator-from-a-distance — "Nobody designed this", "People tend to..." → Put the reader in the room. Use "you".

Passive voice — Always find the actor. Put them at the front.

Scoring

Rate 1-10 on each dimension:

Dimension	Question
Directness	Statements or announcements?
Rhythm	Varied or metronomic?
Trust	Respects reader intelligence?
Authenticity	Sounds human?
Density	Anything cuttable?

Below 35/50: revise.

If 5+ of the 24 patterns are present: very likely AI-generated. If 10+ patterns: almost certainly AI-generated.

Adding Personality

Use references/personality-injection.md for the full guide. Quick version:

• Parenthetical asides — (honestly, this part gets me every time) — 1-3 per 500 words max
• Tangents — "Speaking of which...", "That reminds me..." — 1-2 per 1000+ word piece
• Random thoughts — "I keep coming back to this:", "Honestly didn't think this would work but..."
• Opinions — React to facts. Don't just report them.
• Acknowledge complexity — "I genuinely don't know how to feel about this"
• Let mess in — Perfect structure feels algorithmic

Pre-Delivery Checklist

Before handing over any draft:

• Any adverbs? Kill them.
• Passive voice? Find the actor, make them the subject.
• Inanimate thing doing a human verb? Name the person.
• Sentence starts with Wh- word? Restructure.
• "Here's what/this/that" construction? Cut to the point.
• "Not X, it's Y" contrast? State Y directly.
• Three consecutive sentences match in length? Break one.
• Paragraph ends with punchy one-liner? Vary it.
• Em dash anywhere? Remove it.
• Vague declarative ("The implications are significant")? Name the specific implication.
• Narrator-from-a-distance? Put the reader in the scene.
• Meta-joiners ("The rest of this essay...")? Delete. Let it move.
• Tier 1 vocabulary word? Remove.
• Chatbot artifact? Remove.
• Generic conclusion? Replace with one specific fact.

Filler Replacements (Fast Reference)

Before	After
In order to achieve this	To achieve this
Due to the fact that	Because
At this point in time	Now
In the event that	If
Has the ability to	Can
It is important to note that	(just say it)
For the purpose of	To
In spite of the fact that	Although
Moving forward	Next / From now
Navigate (challenges)	Handle, address
Lean into	Accept, embrace
Deep dive	Analysis, examination
Take a step back	Reconsider
Circle back	Return to
Game-changer	Significant, important

Troubleshooting

Still sounds robotic after fixing patterns? You removed AI tells but didn't add personality. Read references/personality-injection.md.

Too casual after humanization? Match personality injection to context. Fewer asides/tangents in formal writing.

Too perfect? Add imperfection: vary sentence length, include a tangent, acknowledge uncertainty, drop a specific detail that feels slightly off-script.

Word feels suspicious but not on the list? Ask: "Would I say this in conversation?" If no, cut it or simplify.