Claude Humanizer Skill: Make German AI Text Sound Human

Humanizer (Deutsch) is a free, open-source Claude skill for people who draft German text with Claude or Claude Code and do not want the result to sound like machine-polished translation paste.

/plugin marketplace add marmbiz/humanizer-de
/plugin install humanizer-de@humanizer-de

git clone https://github.com/marmbiz/humanizer-de ~/.claude/skills/humanizer-de

Use it when a German draft is useful but too smooth, too generic, too evenly polished, or not quite German in the way a real editor would write it. The skill rewrites toward clearer rhythm, cleaner claims, and a voice that fits the audience.

The mechanism is still precise: Humanizer (Deutsch) checks 66 German-specific AI-writing patterns across language, structure, evidence, typography, rhythm, and register. But pattern detection is not the product. It is the map for a controlled rewrite that keeps facts, claim direction, register, tone, and proportion intact.

Version 5.0.0 adds a compact one-command audit workflow. The QGIR safeguard, introduced in v4.1.0, still matters: a second revision round only starts when a minimal first pass leaves real HIGH or MEDIUM issue clusters, and it stops when more editing would over-polish the text, weaken the facts, distort the tone, inflate the copy, or exceed the edit budget.

What the Humanizer Does

You call it with /humanizer or just say "Humanize this text for me."

Humanizer (Deutsch) reviews German text for 66 AI-writing patterns across 10 categories. It looks for problems such as generic transitions, inflated claims, unnatural German rhythm, too-smooth paragraph flow, register drift, vague authority language, weak evidence handling, and over-polished formulations.

The skill does not simply rewrite everything. It is designed to preserve what should stay stable: factual anchors, claim direction, register, tone, and proportion.

Depending on the workflow, you typically get:

1. the selected mode
2. the main patterns found
3. the specific passages changed
4. a short quality, register, and factuality audit

When used inside a repo workflow, it can also make direct file edits and summarize what changed.

QGIR (since v4.1.0). If the first minimal pass solves the real problems, the process stops. If meaningful HIGH or MEDIUM clusters remain, QGIR can revise again within a strict edit and pass budget.

New in v3: Voice calibration. Provide a sample of your own writing and the skill analyzes your sentence rhythm, word choices, and quirks — then applies them to the rewrite. Instead of generic "clean" output, you get text that sounds like you.

Three modes adjust the correction to your context:

Default is Neutral when the context is unclear.

Recently added (details in the changelog below):

• v5.0 – one audit command: a new combined check runs Unicode, rhythm, pattern, and register linting in a single call and prints compact output. Faster in daily use, same substance.
• v4.3 – factual reliability sharpened: fabricated or unverifiable sources now count as a HIGH evidence problem rather than a style issue; speaker stance (I/we/one) is derived only from input and target profile, never invented.
• v4.2 – pattern 66: the fake-analysis appendix ("…which underscores X"), a pseudo-analytical relative clause that adds no information. Brings the catalog to 66 patterns.
• v4.1 – QGIR and QGIR-Stop: Quality-Guided Iterative Revision adds a controlled second revision round when a minimal first pass leaves real HIGH/MEDIUM clusters. QGIR protects edit budget, factual anchors, register, claim direction, and unwanted expansion.
• v4.0 – standalone project: The Humanizer now follows its own versioning scheme without the fork suffix — its own roadmap instead of upstream tracking. Plus two new patterns: AI marker vocabulary (the German counterparts to "delve" and "tapestry") and copula avoidance ("fungiert als" instead of "ist").
• v3.8 – six new patterns, five-pass workflow, and a rhythm linter: abstract-noun stacking, fabricated first-person anecdotes, synonym rotation, isometric documents, markerless closure compulsion, and modal-particle anomalies. The cleanup now runs in five fixed passes (artifacts → lexis → structure → rhythm → self-audit), and a new measurement script delivers deterministic rhythm metrics instead of gut feeling.
• v3.7 – two new patterns and plugin install: aphoristic empty formulas ("X is the language of Y") and decorative Markdown structure (single-row tables, skipped heading levels, thematic breaks before headings). The skill now also installs directly from the Claude Code plugin marketplace.
• v3.6 – realistic about detectors: two new patterns (colon-title scheme, uniform sentence rhythm) and a clear stance on what online AI detectors actually measure — and what you therefore should not mangle in your text just to chase a score.
• v3.5 – leaner architecture: the pattern catalog, decision tables, and a dedicated Unicode/quote linter are split out; the skill loads only what it needs.

Severity ranking (HIGH / MEDIUM / LOW) for each pattern lets you focus on what matters most. HIGH patterns are almost always AI. LOW patterns only stand out when they cluster.

Why German AI Text Is Different

English and German diverge in their vulnerabilities to LLMs. The same model that produces flawless English can betray itself immediately in German through patterns that native English speakers don't notice.

Take these examples:

• Participle-I constructions like "gewährleistend" or "hervorhebend" (ensuring, highlighting). In English, "-ing" forms are natural everywhere. In German, this construction screams LLM.
• Overused transition phrases like "Darüber hinaus" (furthermore) appearing three times per paragraph. Native German writers vary their transitions. LLMs repeat the same mechanical connectors.
• Em-dashes everywhere — a punctuation habit from English that German doesn't share natively.
• Vague authorities like "Experten sagen" (experts say) with no sources attached.
• Symbolic overload like "steht als Zeugnis für" (stands as testimony to) — nobody writes like this.
• Promotional tone with "atemberaubend" (breathtaking) in contexts where it doesn't belong.
• Chatbot artifacts like "Stand Januar 2024" (as of January 2024) appearing in articles written months later.

Before (LLM):

Die atemberaubende Stadt mit ihrem reichen kulturellen Erbe steht als Zeugnis für die künstlerische Brillanz vergangener Generationen.

"The breathtaking city with its rich cultural heritage stands as testimony to the artistic brilliance of past generations."

After (human):

Die Stadt hat eine lange Geschichte. Ihre Denkmäler zeigen die Handwerkskunst des Mittelalters.

"The city has a long history. Its monuments show medieval craftsmanship."

Less decoration, more substance.

66 Patterns in 10 Categories

The Humanizer uses pattern checks across ten categories:

1. Language & Tone (18 patterns, mostly HIGH)

Symbolic overload, promotional language, editorial comments, mechanical conjunctions, section summaries, participle-I constructions, vague authorities, forced conclusions, negative parallelisms (now including clipped negation fragments like "kein Raten.", "keine Kompromisse."), tricolon overuse, false extensions, misplaced "Fazit" sections, abstract-noun stacking ("verschiedene Maßnahmen" instead of the concrete thing), synonym rotation for the same entity ("die Hansestadt", "die Elbmetropole"), modal-particle anomalies (close-register German with zero "ja", "eben", "wohl" — or far too many), AI marker vocabulary ("beleuchten", "eintauchen", "spannend", "die digitale Landschaft" — the German counterparts to "delve" and "tapestry"), copula avoidance ("fungiert als", "verfügt über" instead of plain "ist"/"hat"), and the fake-analysis appendix ("…was X unterstreicht" — a pseudo-analytical relative clause that adds no information).

2. Style (4 patterns, MEDIUM/LOW)

Excessive bold text, false lists, emojis before headings, em-dash overuse (now with a replacement hierarchy: period > comma > colon > semicolon > parentheses > rephrase, plus detection of paired inserts and dash variants).

3. Communication (6 patterns, all HIGH)

Letter-style writing, collaborative chatbot phrases ("I hope this helps!"), knowledge cutoff references, prompt refusals, placeholder text, links to search queries.

4. Markup (6 patterns, mostly MEDIUM)

Markdown instead of wikitext, broken wikitext and AI tool artifacts (oaicite tags, contentReference spans, turn0search0 references), dead links, full citation fabrication (now HIGH since v4.3.0: hallucinated publications, non-existent journals, utm_source parameters), incorrect reference formats, wrong categories.

5. Miscellaneous (3 patterns, LOW/MEDIUM)

Abrupt cutoffs, style shifts mid-text, first-person edit summaries.

6. Rhetoric & Structure (11 patterns)

7. Argumentation & Evidence (5 patterns)

LLMs hide the actor behind passive voice and subjectless sentences. They stack conditionals where a direct statement would do. Most telling: the swing between over-assertion ("revolutionary", "without doubt") and over-hedging ("seems possibly", "could perhaps") within the same paragraph. When sources are missing, they often add another tell: plausible filler where the text should simply say the point is not documented.

Fabricated first-person experience is the second-order tell: it often appears precisely when someone tries to make AI text sound "more human." Staged anecdotes and forced casualness ("Ehrlich gesagt", "Keine Sorge") are fabrication, not style. That's why the Humanizer never invents experience when rewriting — voice comes only from your writing sample or facts you explicitly provide.

Patterns 32–34 were adapted from upstream PR #39. Patterns 35–38 were adapted from upstream PR #67. Patterns 39–41 are from v3.1, adapted from upstream PRs #79, #80, #84, #85, #94, #96. All have German-specific phrasing and examples.

8. Additions (4 patterns, new in v3.2)

Four patterns drawn from the German Wikipedia's Erkennung KI-Einsatz guideline and its Schnelltest KI companion:

Source incongruence is particularly tricky: the source exists, the DOI validates, the author did publish — only the paper doesn't actually support the claim. A classic LLM hallucination pattern that simple fact-checking tools miss. False friends like "eventuell" (eventually = finally, not "maybe") are corrected regardless of mode because they are semantic errors.

9. Typography & Format (7 patterns)

This category was added in v3.3. It catches texts that are convincing in substance but give themselves away through typographic anglicisms or decorative formatting.

The quotation-mark problem is unusually stubborn: Claude picks the wrong closing German quote systematically, and prompting alone won't reliably fix it. The Humanizer flags those spots — the actual fix belongs in a post-processor or linter. Not every odd quote is a tell, though: the only hard AI signal is the asymmetry — a German opening quote (U+201E) paired with a wrong or straight closing mark instead of the correct U+201C. Consistently straight quotes, by contrast, are usually a CMS or editor artifact, not an AI tell; consistently English curly quotes are a weak signal at best. Treating every straight quote as AI just manufactures false positives.

Obsessive parataxis is the opposite kind of tell: subtle. Each individual sentence is correct, readability scores fine — but the monotony betrays the machine. The fix isn't "rewrite everything", it's turning every third sentence into a complex sentence with subordination. Exception: if staccato is the intended style (advertising, manifestos), the "don't touch" rule for soft-pattern clustering applies.

Before (LLM):

Das Team Analysierte Die Daten. Die Ergebnisse waren eindeutig. Die Conversion stieg um 3.5 Prozent. Das Projekt wurde im Budget abgeschlossen.

After (human):

Das Team analysierte die Daten und kam zu einem eindeutigen Ergebnis: Die Conversion stieg um 3,5 Prozent, obwohl das Projekt im Budget blieb.

10. Title & Sentence Structure (2 patterns, new in v3.6)

Two patterns that only stand out when they cluster — and the only ones in this catalog that statistical detectors also measure (more on that below).

What Should Stay Stable

The Humanizer has a guardrail against over-editing. Not every polished or formally correct piece of writing is a problem.

These signals should stay untouched on their own:

• perfect grammar and consistent style
• one dash or curly quotes
• dry prose without specific patterns
• one transition word such as "allerdings" or "zudem"
• unsourced claims without additional source or speculation patterns

The skill also preserves positive human signals: concrete details, unresolved tension, era-bound references, genuine asides, self-corrections, and varied sentence length. The point is to improve the text without flattening the author.

What About AI Detectors?

AI detectors can be a useful signal, especially when they point to visible writing problems: monotonous rhythm, generic phrasing, empty transitions, or copy that feels too evenly polished.

But a detector score is not the same as quality.

Many tools estimate two statistical quantities:

• Perplexity — how predictable the next word is. Precise, smooth technical prose is highly predictable and scores low perplexity.
• Burstiness — how much sentence length and structure vary. Uniform sentences produce low burstiness.

The pretty labels such tools attach — "mechanical precision," "impersonal tone," "robotic formality" — are often translations of those numbers. A German text can score badly and still be accurate, useful, and appropriate. It can also score well and still sound generic, vague, or unconvincing.

That is why Humanizer (Deutsch) does not treat detector context as an automatic error. The better question is editorial: does the text work for real readers? Does it keep the facts stable? Does the register fit? Does the voice sound credible? Are the claims supported?

If detector feedback reveals a real writing problem, the Humanizer can help fix the writing problem:

• Colon-title scheme (pattern 54): When the H1, the caption, and several subheadings all follow the "Keyword: explanatory tail" shape, the result is a mechanical rhythm. A single colon title is perfectly fine — the clustering is the signal.
• Uniform sentence rhythm (pattern 55): When nearly every sentence is the same length and starts with the subject, the text turns monotone. The fix is not to insert errors but to deliberately spread sentence length — a short sentence next to a long, structured one.
• Abstract-noun stacking (pattern 58): "Verschiedene Maßnahmen zur Verbesserung der Verkehrssituation" ("various measures to improve the traffic situation") becomes stronger when the text names the concrete measures.

Since v4.1.0, this stance is also reflected in the workflow: detector context is context, not an automatic contract violation. QGIR revises again only where real quality problems remain, and stops when additional edits would make the text weaker.

Why I Created the German Humanizer

I discovered Siqi Chen's original Humanizer and immediately saw the gap: it worked brilliantly for English, but German AI had different patterns. Testing it on German text was like using an English spell-checker on German — not wrong, just missing the point.

The German Wikipedia maintains its own guide to AI-generated content indicators. The English Wikipedia has a comparable resource. Siqi's original pulls from the English one; the German version documents something different. I used both as the foundation.

The philosophy is the same as Siqi's tool — analysis, not blind rewriting. But the patterns are German-specific. Since v4.0.0 the project stands on its own: it follows its own versioning scheme and roadmap, roughly half of its 66 patterns have no upstream counterpart, and it ships deterministic linters and a test suite the original doesn't have. Since v4.1.0, QGIR adds a controlled second revision round with stop rules against over-editing, and since v5.0.0 a single audit command bundles all linters. The original remains both the inspiration and a source of ideas worth adapting.

Who Needs This

• German content creators using AI who still want the text to sound like them
• Marketing teams improving German landing pages, LinkedIn posts, help articles, newsletters, or product copy
• Teams working with factual German text where claims, sources, tone, and register must stay stable
• Wikipedia editors evaluating German submissions
• Bilingual teams where English-speaking editors need to understand German AI-writing patterns
• Anyone turning useful German AI drafts into better German copy

Credits and Open Source

The tool is MIT licensed and open source. It builds on:

• Pattern research: German Wikipedia's AI detection guide + English Wikipedia's AI writing signs
• Original concept & English Humanizer: Siqi Chen (blader)
• German adaptation: github.com/marmbiz/humanizer-de

I built the German version. Siqi built the original. Both Wikipedias documented the patterns.

Changelog

This post was written with AI assistance and reviewed with language-specific awareness. That is the real value of a humanizer: not making AI invisible, but making the writing worth reading.

More from the lab

• AI Overviews steal clicks: How GEO works: what AI answers do to content visibility.
• Profile: who is behind this lab.