Skip to content

LeoYeAI/agent-master-skills / skills/ai-humanizer

humanizer

Verified

Humanize AI-generated text by detecting and removing patterns typical of LLM output. Rewrites text to sound natural, specific, and human. Uses 24 pattern detectors, 500+ AI vocabulary terms across 3 tiers, and statistical analysis (burstiness, type-token ratio, readability) for comprehensive detection. Use when asked to humanize text, de-AI writing, make content sound more natural/human, review writing for AI patterns, score text for AI detection, or improve AI-generated drafts. Covers content, language, style, communication, and filler categories.

FileKindSize
README.mdmarkdown15.8 KB
SKILL.mdskill md7.8 KB
assets/banner.mdmarkdown331 B
docs/CONTRIBUTING.mdmarkdown2.0 KB
docs/EXAMPLES.mdmarkdown5.3 KB
docs/PATTERNS.mdmarkdown2.4 KB
eslint.config.jsscript559 B
package.jsonmanifest1.3 KB
references/ai-vocabulary.mdmarkdown5.5 KB
references/patterns.mdmarkdown10.0 KB
references/style-guide.mdmarkdown3.9 KB
scripts/analyze.shscript301 B
scripts/humanize.shscript315 B
src/analyzer.jsscript14.8 KB
src/cli.jsscript19.6 KB
src/humanizer.jsscript14.2 KB
src/patterns.jsscript28.1 KB
src/stats.jsscript9.5 KB
src/vocabulary.jsscript15.1 KB
tests/analyzer.test.jsscript13.2 KB
tests/calibration.test.jsscript7.2 KB
tests/edge-cases.test.jsscript7.1 KB
tests/fixtures/ai-sample-1.txtmarkdown2.2 KB
tests/fixtures/ai-sample-2.txtmarkdown1.1 KB
tests/fixtures/human-sample-1.txtmarkdown1.1 KB
tests/humanizer.test.jsscript6.4 KB
tests/performance.test.jsscript3.2 KB
tests/statistics.test.jsscript7.9 KB
vitest.config.jsscript152 B

references/style-guide.md

View on GitHub
# Style guide — how to write like a human

Removing AI patterns is half the job. The other half is writing with personality. Sterile, voiceless text is just as obvious as slop.

## The human test

Read your text aloud. If it sounds like a press release, a Wikipedia article, or a customer service chatbot, it needs work. Good writing sounds like a specific person thinking on paper.

## Have opinions

Don't just report facts — react to them.

**Soulless:** The experiment produced interesting results. Some developers were impressed while others were skeptical.

**Human:** I genuinely don't know how to feel about this one. Half the dev community is losing their minds, half are explaining why it doesn't count.

## Vary your rhythm

Short punchy sentences. Then longer ones that take their time getting where they're going, maybe with a clause or two along the way. Mix it up.

**Monotone:** The tool works well. It processes data quickly. Users report satisfaction. The interface is clean.

**Varied:** The tool works well. It's fast — embarrassingly fast, actually, compared to the manual process we'd been limping along with for years. Users seem happy with it, though I suspect half of them haven't tried the advanced features yet.

## Acknowledge complexity

Real humans have mixed feelings. "This is impressive but also kind of unsettling" is more honest than either pure praise or pure criticism.

**AI-flat:** This is an innovative solution that addresses key challenges.

**Human:** It's clever, but it makes me nervous. The failure modes aren't well understood.

## Use first person when it fits

"I" isn't unprofessional — it's honest.

- "I keep coming back to..."
- "Here's what gets me..."
- "I've seen this go wrong when..."
- "What I don't understand is..."

## Let some mess in

Perfect structure feels algorithmic. Tangents, asides, and half-formed thoughts are human. Not every piece needs them, but they signal a real person behind the words.

## Be specific

"This is concerning" tells you nothing. "There's something unsettling about autonomous agents running at 3am while nobody watches" tells you everything.

Replace vague adjectives with concrete details:
- "Important city" → "Portland, population 650,000"
- "Recent study" → "A 2024 Stanford study of 500 developers"
- "Significant growth" → "Revenue grew 34% in Q3"

## Use simple verbs

| Instead of | Use |
|------------|-----|
| serves as | is |
| features | has |
| boasts | has |
| utilizes | uses |
| facilitates | helps |
| encompasses | includes |
| leverages | uses |
| garners | gets |

## Cut filler words

Every word should earn its place. If you can remove a word without changing the meaning, remove it.

| Remove | Keep |
|--------|------|
| In order to succeed | To succeed |
| Due to the fact that | Because |
| It is important to note that the data shows | The data shows |
| At this point in time | Now |
| Has the ability to process | Can process |

## One qualifier is enough

"Could potentially possibly" means the same as "could." Pick one qualifier and commit.

**Over-hedged:** It could potentially be argued that this might possibly have some positive effect.

**Direct:** The policy may help.

## End with something real

Don't reach for a platitude. End with a fact, a question, or nothing at all.

**Generic:** The future looks bright. Exciting times lie ahead.

**Specific:** The company plans to open two more locations next year. Whether that's ambitious or reckless depends on the Q4 numbers.

**Or just stop:** The company plans to open two more locations next year.

## Formatting guidelines

- Use sentence case for headings (capitalize first word and proper nouns only)
- Avoid bold emphasis in body text — let the writing carry the weight
- Don't decorate lists with emojis
- Use straight quotes, not curly quotes
- Prefer commas and periods over em dashes
- Don't use inline-header lists ("**Topic:** description of topic")
- If you need a list, keep items brief and parallel
receiving-code-reviewobra/superpowersUse when receiving code review feedback, before implementing suggestions, especially if feedback seems unclear or technically questionable - requires technical rigor and verification, not performative agreement or blind implementationStars: 322kUpdated Claude CodeCodexVerifiedInstall
requesting-code-reviewobra/superpowersUse when completing tasks, implementing major features, or before merging to verify work meets requirementsStars: 322kUpdated Claude CodeCodexVerifiedInstall
using-git-worktreesobra/superpowersUse when starting feature work that needs isolation from current workspace or before executing implementation plans - ensures an isolated workspace exists via native tools or git worktree fallbackStars: 322kUpdated Claude CodeCodexVerifiedInstall
verification-before-completionobra/superpowersUse when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions alwaysStars: 322kUpdated Claude CodeCodexVerifiedInstall