linkedin-humanizer
Remove AI tells from LinkedIn posts/comments: 2026 vocabulary density, reveal bridges, staccato fragments, stacked triads, performed sincerity. Tiered rewriter plus `--mode audit` LinkedIn post auditor and `--mode profile` voice builder.
Install / Use
npx skills add sergebulaev/linkedin-skills --skill linkedin-humanizerInstalls into whichever agent you are using.
SKILL.md
Installable skill definition
Quality Score
Category
MarketingSupported Platforms
Our assessment of linkedin-humanizer
linkedin-humanizer scores 92/100 on our quality scale, 180th of 642 Marketing skills we index (top 29%).
Its SKILL.md is 19 KB long, well organised into 29 sections with 1 code example: a thorough specification that gives an agent plenty to work with.
With 5,507 GitHub stars, it is one of the more widely adopted skills in the catalogue.
Maintenance, license and trust
- The repository was last updated 20 days ago, so linkedin-humanizer is actively maintained.
- Our last check on 2026-10-04 found the source still online.
- It is released under the MIT license, a permissive license that allows use, modification and commercial use with attribution.
- Its trust signals score 100/100, with no cautions. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.
Safety scan
No issues foundOur scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful.
AI review by kimi-k2.7-code on 2026-09-29. Automated pattern scan on 2026-09-29. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.
linkedin-humanizer compared with similar skills
All 4 of these similar skills score higher than linkedin-humanizer; compare them before choosing.
| Skill | Score | Stars | Updated | Format |
|---|---|---|---|---|
| linkedin-humanizer (this skill)by sergebulaev | 92 | 5.5k | 20d ago | SKILL.md |
| LocalAIby mudler | 100 | 50.4k | today | MCP Server |
| x402by browser-use | 100 | 121.0k | 3d ago | SKILL.md |
| algorithmic-artby anthropics | 100 | 177.9k | 55d ago | SKILL.md |
| pptxby anthropics | 100 | 177.9k | 55d ago | SKILL.md |
Frequently asked questions
- How do I install linkedin-humanizer?
- Run
npx skills add sergebulaev/linkedin-skills --skill linkedin-humanizer. The install tabs above show the steps for each supported agent. - Which AI agents does linkedin-humanizer work with?
- It is written for Universal, as a SKILL.md file. Other agents that read the same format can often use it too.
- Is linkedin-humanizer safe to use?
- Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful. It is MIT-licensed and scores 100/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
- Is linkedin-humanizer still maintained?
- The repository was last updated 20 days ago, so linkedin-humanizer is actively maintained.
Skill content
View source on GitHubname: linkedin-humanizer
description: "Remove the AI tells human readers and LinkedIn's AI-slop filter react to in a post or comment: 2026 vocabulary by paragraph density, reveal bridges, staccato fragments, stacked triads, performed sincerity. Tiered rewriter (forensic / strict / aesthetic / all) plus --mode audit pass-fail review and --mode profile voice profile builder. Not for beating AI detectors (no edit reliably does). Keywords: humanize, de-AI, reads like ChatGPT, AI slop, scrub AI tells, review this draft, audit before posting."
LinkedIn Humanizer V3
Rewrites any text to remove the AI tells that human readers notice and that LinkedIn's "AI slop" filter reacts to. Based on Wikipedia's "Signs of AI writing" taxonomy, the 2025-2026 stylometry literature, and our own length-controlled corpus. V3 (2026-09): recalibrated on 2026 evidence. Vocabulary is scored by density, em dashes are capped instead of banned, forced rhythm is now a tell instead of a fix, and there is an over-correction guard.
What this skill does not do: it does not make text "pass" GPTZero, Pangram, Turnitin or Originality. Those are trained classifiers keyed on the instruction-tuning style signature; prompt-style "sound like a real person" rewrites are caught 92-95% of the time (VUB IJEI 2026, Russell 2025), and light mechanical rewriting raises detectability (arXiv 2603.17522). No post-hoc edit reliably beats a Pangram-class detector, and detector scores on LinkedIn-length text (100-300 words) are noise. The real value is elsewhere: expert human readers cite vocabulary (53%) and sentence structure (36%) as what gives AI text away, and LinkedIn's July 2026 slop-report button costs a flagged post roughly 40% of its views. This skill removes what those readers and that filter react to.
What changed in V3
Evidence tier in brackets: [strong] = replicated across 2+ independent 2025-2026 studies or our own length-controlled corpus; [vendor] = single platform or vendor dataset; [weak] = one study or expert-panel report.
- Vocabulary moved from a delete-list to density scoring. The 2023-24 words (delve, tapestry, realm, journey) are decaying as humans avoid them [strong: Geng & Trotta 2025]. The durable 2026 markers are common words (significant, crucial, notably, comprehensive, insights, robust, leverage, foster, landscape, nuanced, streamline, elevate) plus grammar: nominalisations and "-ing" clause openers at 5.3x human rate [strong: Kobak Sci Adv 2025; Wu et al 2026; PNAS 2025]. AI vocabulary is also the one marker consistently reach-negative on LinkedIn in our own corpus (0.74-0.84 author-relative) [strong]. One marker in a paragraph is not a verdict. Three or more is.
- Em dash is no longer a tell. GPT-5.4 emits 1.43 per 1,000 words, below the 3.23 human baseline; 29% of human captions and 23% of top-creator LinkedIn posts in our corpus use one (author-relative ratio 1.09) [strong]. Zero em dashes is now its own tell (the writer is trying to look human). New rule: cap at about 1 per 100 words, replace excess with comma, colon, parentheses or a rewrite. Never a period.
- Forced burstiness is the #1 2026 tell, not the fix. LLM sentence-length variance is half of human [strong], but detectors do not score it, mechanical long/short alternation is a learnable humanizer fingerprint [weak: DAMAGE 2025], and on LinkedIn sentence-length variance is not an engagement lever in either direction (our corpus, n=397, within-creator: null to slightly negative) [strong]. "Short. Punchy. Done.", "No X. No Y. Just Z.", one-word paragraphs and "The result?" reveals are the current top tells. Pass 2 is now RHYTHM, not BREAK: fix machine-flat rhythm, never manufacture variance.
- Rule of three is still a tell, at density. Tricolon runs at 2x expert-human rate across 2026 frontier models [strong: arXiv 2604.19768]. Stacked, perfectly parallel triads and 3+ per post get scrubbed. One natural triple stays (26% of top human tweets have one).
- Fingerprint injection was half wrong. Named entities and concreteness are supported [strong: lower entity density in LLM text across 3 studies]; an odd-precision number with a referent in line 1 lifts likes 34% [vendor]. Bare numbers are not a discriminator, and inserted hedges and confessions backfire: performed hesitancy is 2x more common in LLM text than expert human text, and sincerity announcements ("let me be honest") are a named 2026 tell [strong: tropes.fyi false vulnerability; Schilke & Reimann 2025]. Pass 3 now asks for a flat, dated, uncomfortable fact instead.
- Over-correction guard. Humanizer output has its own fingerprint; "writing slightly worse on purpose" now reads as a tell [weak: DAMAGE 2025; slopotron]. Pass 4 checks whether Passes 1-3 introduced the very patterns they were meant to remove. Edits are proportional to real problems. When in doubt, leave it.
See sub-skills/rules-explainer.md for per-rule justification, defenses, and citations, and references/tier-rationale.md §V3 for the evidence.
When to use
- Before publishing any AI-drafted post or comment (rewrite mode)
- Pre-publish review of a finished draft (audit mode, see
sub-skills/post-audit.md) - When a draft feels off and you can't pinpoint why
Input
Any text (post, comment, reply, DM). Optional: target voice samples (past human posts by the user).
Output
- Rewritten text with AI tells removed
- Diff showing what changed and why
- Per-paragraph tell density (markers per paragraph; 3+ triggered a rewrite)
- Reader-read confidence: "reads human", "mixed", "reads AI" (this is a reader-tell estimate, not a detector score)
- Tier applied (which mode was used)
Modes
# Default: forensic + strict (recommended for LinkedIn)
linkedin-humanizer <text>
# Forensic only: minimum-touch, just kill the leakage
linkedin-humanizer --mode forensic <text>
# Strict: forensic + density-scored 2026 vocabulary, reveal bridges, staccato (the LinkedIn-default config)
linkedin-humanizer --mode strict <text>
# Aesthetic: strict + style rules (single natural triads, passive voice, defendable vocab)
# Use when target audience is Wikipedia editors / academic readers / AI-tell hunters
linkedin-humanizer --mode aesthetic <text>
# All: every rule. Maximum scrub. Will flatten literary writing and trip the Pass 4 guard.
linkedin-humanizer --mode all <text>
# Audit: detection-only pass-fail review. No rewrite.
# Runs the 2026 algorithm checklist: length, hook, CTA, structure, AI tells.
# Returns Blockers + Warnings + suggested fixes. See sub-skills/post-audit.md.
linkedin-humanizer --mode audit <text>
# Profile: build/update the user's Voice & Brand Profile so every writing
# skill drafts in their real voice. Learns from 3-6 pasted posts (portable, no
# token) or, if APIFY_TOKEN is set, from pulled activity. Writes
# ../../references/voice-profile.md. See sub-skills/voice-profile.md.
linkedin-humanizer --mode profile
The four passes
Pass 1: SCRUB (score, then delete or replace)
The scrub pass applies tiered catalogs to delete or replace AI tells. The unit of judgement is the paragraph, not the word: count markers per paragraph, rewrite the paragraph at 3+, leave a single marker alone unless it is a reveal bridge or forensic leakage. Full regex source, replacement maps, and detection functions live in references/scrub-rules.md; load that file when actually executing the scrub.
FORENSIC tier (always on): real model leakage no human produces. Covers AI tool markers (oaicite, contentReference, turn0search0, attached_file, grok_card), knowledge-cutoff disclaimers ("As of my last update..."), phrasal templates ([Your Name], 2025-XX-XX), em dash density above 1 per 100 words, and outline-formula closers ("Despite its X... Looking ahead...").
STRICT tier (default on): what readers and the slop filter react to. Covers punctuation normalization (curly to straight quotes, -- to a comma or rewrite; excess em dashes to comma, colon or parentheses, never a period), the durable 2026 vocabulary set scored by density (significant, crucial, notably, particularly, comprehensive, insights, robust, leverage, foster, landscape, nuanced, multifaceted, holistic, streamline, elevate, empower), grammatical markers (nominalisations, sentence-opening "-ing" clauses), the 2026 LinkedIn layer (quietly, matters, compound, signal, "the work", "built different", load-bearing, "doing the heavy lifting", "let that sink in", "that's the real story"), reveal bridges measured reach-negative ("The result?" -4.8%, "It's not X, it's Y" -4.9%, "Stop X, start Y" -6.7%, "Here's what/how" -4.3%), all 6 forms of negative parallelism, stacked or perfectly parallel triads and any 3rd triad in a post, and cliché closer tells ("What do you think?", "Tag someone who needs this").
AESTHETIC tier (opt-in only, will flatten literary writing): patterns AI uses but humans use legitimately. Covers the one remaining natural triad, decaying 2023-24 vocabulary that is now mostly harmless (delve, tapestry, realm, intricate, journey, paradigm), defendable normal English (cultivate, vibrant, garner, showcase, underscore), and passive voice (academic-writing defense ignored).
Pass 2: RHYTHM (restore natural variance)
Detectors do not score burstiness, and on LinkedIn sentence-length variance is not an engagement lever in either direction. What readers do notice is the mechanical-uniformity tell (every sentence the same length, machine-flat; structure is 36% of expert judgments) and, worse, the staged variance that second-generation humanizers add. So Pass 2 has two jobs: fix rhythm only where it reads machine-flat, and remove manufactured variance everywhere. It never adds variance as a tactic.
- Per paragraph: one genuinely long sentence (25+ words, with a subordinate clause that does real work) next to a short one is fine and is what human variance looks like. Two or three mid-length sentences in a row are also fine. Edit only when every sentence in the paragraph runs the same length and reads flat, and then edit one sentence, not the paragraph.
- Standalone fragments: at most 2 per post, total. "Worth it." once is a voice quirk. Three in a post is a pattern.
- Banned outright (rewrite as full sentences): "The X? Y." reveals; "No X. No Y. Just Z."; "All the X. None of the Y."; "Simple. Effective. Easy." adjective stacks; one-word paragraphs ("Still." "Mostly." "Exactly."); pseudo-Socratic Q&A ("Why? Because..."); "Short. Punchy. Done." staccato runs. Fragment runs are the tell.
- Layout is not rhythm. One or two sentences per paragraph with blank lines between them is mobile-native LinkedIn formatting and stays (our corpus shows a mild uniform-rhythm advantage for that one-idea-per-line format at 112-204 words). Fragment-for-drama inside those paragraphs is the tell. Keep the layout, fix the sentences.
- Length note: on LinkedIn our corpus (n=397, author-normalised) shows sentence-length variance is not an engagement lever (null to slightly negative within-creator, no length-dependent flip). The short-form "don't force variance" rule applies to sibling platforms (Threads, short X); here it applies at every length.
- Break perfect parallel structures with one asymmetric sentence, once. Never alternate long/short/long/short across a post; that seesaw is the humanizer fingerprint.
Target: Flesch reading ease >55. No sentence-length variance target. The check is "does any paragraph read machine-flat, and did I add a staccato pattern," not a number.
Pass 3: ADD (human fingerprints)
Require at least:
- One odd-precision number WITH a named referent: who, what, when, or what it cost ("$4,730 in Vercel overages, March invoice", not "$5k" and not "significant costs"). A bare number is not a fingerprint; LLM news copy uses more numbers than humans do. The referent is what carries the signal.
- One named entity (real person, company, date, city, tool)
- One first-person
Truncated for display — read the full file on GitHub.
Related Skills
LocalAI
50.4kLocalAI is the open-source AI engine. Run any model - LLMs, vision, voice, image, video - on any hardware. No GPU required.
x402
121.0kSet up Browser Use Cloud payments with x402 — pay per request from a crypto wallet (USDC on Base mainnet), no signup or API key.
algorithmic-art
177.9kCreating algorithmic art using p5.js with seeded randomness and interactive parameter exploration. Use this when users request creating art using code, generative art, algorithmic art, flow fields, or particle systems.
pptx
177.9kUse this skill any time a .pptx or .potx file is involved in any way — as input, output, or both. This includes: creating slide decks, pitch decks, or presentations; reading, parsing, or extracting text from any .pptx or .potx file (even if the extracted content will be used elsewhere, like in an em…
Languages
Trust signals
From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.
