[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7ES45J1NP5RGARXKK7SPFWW":3,"skill-same:01M7ES45J1NP5RGARXKK7SPFWW":81},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":15,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":27,"commitSha":28,"skillDirectory":29,"forksCount":13,"repositoryHtmlUrl":30,"repositoryHomepageUrl":6,"ownerHtmlUrl":31,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":32,"install":33,"resources":46},"01M7ES45J1NP5RGARXKK7SPFWW","humanizer",null,"Humanize and de-slop AI-sounding prose while preserving meaning, specific detail, and genuine human quirks. When a voice sample or style profile is available, rewrite in that writer's actual voice. Remove recurring AI tells such as inflated significance, stock vocabulary, uniform rhythm, excessive hedging, formulaic structure, and formatting artifacts. Also clean suspicious invisible Unicode and copy-paste residue conservatively, without treating them as proof of authorship. Use whenever a user wants to humanize, de-AI, de-slop, or de-robotify text; fix prose that sounds corporate, generic, like ChatGPT, or otherwise off; match a draft to a writer or named author; edit for authentic voice and rhythm; or clean zero-width characters, exotic spaces, bidi controls, and similar hidden text residue.","mhyahya854\u002FAI-Second-Brain_Universal-Rules","mhyahya854","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F194078115?v=4","01M7EGSP2M9T858SDHJMXQ257Y","01M7EGSP3N6HN6879SVDPMD4DZ",0,"MIT","https:\u002F\u002Fgithub.com\u002Fmhyahya854\u002FAI-Second-Brain_Universal-Rules\u002Fblob\u002Fmain\u002Fskills\u002F_intake\u002Fgithub\u002Fhannsxpeter\u002Fhumanizer\u002Fhumanizer\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Fmhyahya854\u002FAI-Second-Brain_Universal-Rules\u002Fblob\u002Fmain\u002Fskills\u002F_intake\u002Fgithub\u002Fhannsxpeter\u002Fhumanizer\u002Fhumanizer\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: humanizer\ndescription: >-\n  Humanize and de-slop AI-sounding prose while preserving meaning, specific\n  detail, and genuine human quirks. When a voice sample or style profile is\n  available, rewrite in that writer's actual voice. Remove recurring AI tells\n  such as inflated significance, stock vocabulary, uniform rhythm, excessive\n  hedging, formulaic structure, and formatting artifacts. Also clean\n  suspicious invisible Unicode and copy-paste residue conservatively, without\n  treating them as proof of authorship. Use whenever a user wants to humanize,\n  de-AI, de-slop, or de-robotify text; fix prose that sounds corporate,\n  generic, like ChatGPT, or otherwise off; match a draft to a writer or named\n  author; edit for authentic voice and rhythm; or clean zero-width characters,\n  exotic spaces, bidi controls, and similar hidden text residue.\nlicense: MIT\nallowed-tools: Read Write Edit Glob Grep\nmetadata:\n  version: 1.3.1\n---\n\n# Humanizer\n\nTurn AI-flavored or generically smooth prose into writing that reads as the\nwork of a person. When a writer's voice is available, make it read as the work\nof *that* person. The goal is prose a careful human reader would not flag,\nwith its meaning and its human texture intact. The goal is not to beat any\nparticular detector, and this skill names and targets none.\n\nThe workflow also includes conservative text hygiene. It can remove\nsuspicious invisible Unicode and normalize copy-paste spacing in the prose it\nrewrites, while preserving characters that carry real linguistic or display\nmeaning. This is not file-provenance or media-watermark tooling.\n\n## When to use this\n\nUse it when someone pastes a draft and says it sounds like AI, asks to\nhumanize or de-slop text, wants their own draft to sound like them, names an\nauthor to match, or says the writing \"feels off,\" \"sounds corporate,\"\n\"reads like a robot,\" or \"is too generic.\" Use it even when they do not say\nthe word \"humanize.\" It also pairs well after any drafting step where the\noutput came out machine-smooth. By default the rewrite stays measured and\nfaithful; if the user explicitly asks for more voice, edge, or opinion, turn\non Stance mode (see below).\n\nUse the text-hygiene part when prose may contain invisible characters,\nzero-width residue, unusual spaces, or direction controls. Treat hygiene as\nformat cleanup, never as evidence that a person or model wrote the text.\n\nDo not use it to disguise authorship for an assessment or to defeat a\nplagiarism or AI-detection system. See Scope and intended use below; reframe\nsuch requests toward genuine quality and voice.\n\n## Core principle: variance over substitution\n\nThe reason AI prose reads as AI is not its vocabulary. It is its uniformity:\neven sentence lengths, even rhythm, even hedging, the same shapes resolved the\nsame way. Swapping \"delve\" for \"explore\" everywhere does not fix that. It just\nmoves the signature, and a uniform \"humanizer dialect\" is still a machine\nfingerprint, just a different one.\n\nReal humanization comes from genuine structural variance. Sentences of very\ndifferent lengths next to each other. A fragment. Then a long, winding one\nthat takes its time. Mixed syntax. The occasional asymmetry a tidy model would\nsmooth away. You are writing for a sharp human reader, not optimizing a score.\nEvery pass below serves this principle: increase real variance, do not install\na new regularity.\n\n## Step 0: Discover the voice (do this every run, before rewriting)\n\nDecide which mode you are in. Announce it in the output header so a wrong\nguess is instantly correctable.\n\n1. **Explicit input wins.** Did the user paste a writing sample, point to a\n   file, or name a well-known author to match? Use that. Read\n   `references\u002Fvoice-matching.md` and proceed in voice mode.\n2. **Otherwise, look for a profile.** `Glob` the working directory and up to\n   two parent levels, in this priority order:\n   `VOICE.md`, then `STYLE-GUIDE.md`, then\n   `voice-profile.md` \u002F `.yaml` \u002F `.json`, then `.manuscript\u002FSTYLE-GUIDE.md`,\n   then a voice or style section inside `AGENTS.md` or `CLAUDE.md`. One strong\n   match: read it, then read `references\u002Fvoice-matching.md`, proceed in voice\n   mode. Several plausible matches: ask one short question to pick.\n3. **Otherwise, generic mode.** No voice target. De-slop toward natural,\n   varied human prose. Do not invent a persona or a backstory; an invented\n   voice is its own failure. Generic mode skips Pass 1 and starts at Pass 2.\n\nThis discovery is filesystem-generic. It interoperates with Scriveno, Pillars,\nor any project that happens to keep a voice file, without depending on any of\nthem.\n\n## Step 0b: Stance mode (opt-in, off by default)\n\nDefault output is measured and faithful: it reports the source's content in\nclean human prose and does not editorialize. That is the right default,\nbecause the most common failure of livelier humanizers is that \"adding voice\"\nbecomes \"adding content,\" and an invented opinion quietly drags an invented\nfact in behind it.\n\nTurn stance mode **on only when the user explicitly asks for it**: \"give it\nsome voice,\" \"make it punchier,\" \"have a take,\" \"don't keep it so neutral,\"\n\"add some edge or opinion.\" Do not turn it on for an ordinary de-slop request,\nand never in voice mode (a discovered profile already dictates stance; do not\noverride it).\n\nWhen stance mode is on, you may add **attitude toward the existing content**:\na judgment, emphasis, dry understatement, first person as the narrator of the\nsame facts, an honest \"this part is the weak spot,\" acknowledged ambivalence\nabout what the source already says, and freer rhythm. This is the same\nmaterial, said by someone with a point of view.\n\nThe hard line, and the entire reason this mode is safe: **stance is a reaction\nto what is there, never the addition of what is not.** You may judge a stated\nfact; you may not invent the fact, its cause, its mechanism, an event, a named\nperson, a quote, or a lived experience to react to. Test every added clause:\n\"is this an opinion about something the source already states, or a new claim\nabout the world?\" If it is a new claim, it is fabrication wearing a voice, and\nit is exactly the failure this skill exists to prevent. When in doubt, the\nopinion is allowed and the new fact is not.\n\nExample. Source states only: infrastructure cost rose 18% during the\nmigration, temporarily.\n\n- Allowed (stance about the stated fact): \"An 18% jump is a real line item,\n  not a rounding error, and 'temporary' is doing some hopeful work in that\n  sentence.\"\n- Forbidden (invented cause dressed as candor): \"costs rose because we were\n  running the old and new systems in parallel\" (the source never says why;\n  this is a new fact).\n\nStance mode raises fabrication risk, so the Pass 3 self-audit and the Meaning\ncheck apply with extra force, and the output header must show stance is on so\nthe user can dial it back. Stance is a requested change of register, so it\napplies even when the density pre-check picks a light pass; density still\ndecides how hard to hunt for tells.\n\n## Step 0c: Density pre-check (match effort to evidence)\n\nBefore rewriting, skim the whole text once and judge how heavily it is\nAI-marked. Count only the **dead-giveaway** tells, not the weak surface\nsignals: chat-UI contamination (pattern 31), knowledge-cutoff disclaimers\n(29), collaborative chatbot artifacts (28), significance inflation (1),\npromotional language (5), AI-vocabulary clustering (16), sycophantic tone\n(7), generic positive conclusion (13). Estimate roughly how many appear per\n100 words, then pick the pass intensity and announce it in the output header:\n\n- **Low (about 0 to 2 per 100 words): light pass.** The text is human-first\n  (a journal, rough notes, a real person's draft). Fix only the dead-giveaway\n  artifacts. Leave lower-tier patterns, rhythm, and voice alone, and bias hard\n  toward \"Deliberately left alone.\" All-or-nothing rewriting here is exactly\n  what strips the quirks that made the writing sound like a person; that is\n  the worst failure this skill can commit, so when density is low, restraint\n  is the default and edits are the exception.\n- **Medium (about 3 to 5 per 100 words): standard pass.** Mixed authorship.\n  Apply the full catalog, but with the restraint of do-not-flag.md; do not\n  flatten what already works.\n- **High (6 or more per 100 words): full pass.** AI-first text. Apply the\n  full catalog thoroughly and push real structural variance.\n\nThis is calibration, not a scan you report mechanically. Effort should match\nevidence. One exception overrides density: any chat-UI contamination string\n(pattern 31) is decisive on its own and is always removed, whatever the\noverall density, because its presence is near-certain confirmation rather\nthan a weak signal.\n\n## Step 0d: Text hygiene preflight (do this every run)\n\nInspect the source before rewriting so invisible formatting does not survive\nor get confused with prose style. Read `references\u002Ftext-hygiene.md` when the\nsource contains non-ASCII formatting characters, when the user mentions\nhidden characters or text marks, or when the interface reports suspicious\ncodepoints.\n\nRemove only characters whose lack of semantic purpose is clear. Preserve\nscript joiners, direction marks needed by multilingual text, variation\nselectors used for display, and non-breaking spaces that carry layout or\nlocale meaning. Do not normalize confusable letters automatically. Invisible\ncharacters do not count toward the density score and do not establish AI\nauthorship.\n\nThis is a pure-prompt skill, so inspection is limited to characters the host\npreserves and exposes. When the prose is in a file and a search tool is\navailable, also run the codepoint search in `references\u002Ftext-hygiene.md`. If\nyou can neither see the characters directly nor search a file, say\n`not verifiable in this interface` in the output header. Never report the\ntext as universally watermark-free.\n\n## The multi-pass workflow\n\nRun the passes in order. Voice is generative; de-slopping is subtractive. If\nyou strip tells first you get flat neutral prose that is then hard to inflect,\nso establish voice first when you have one.\n\n### Pass 1: Voice injection (voice mode only)\n\nRewrite the draft *toward the target voice* before removing any tells. Set the\ncadence, diction, sentence-length signature, stance, and characteristic moves\ndescribed in `references\u002Fvoice-matching.md`. Match the writer's instincts and\ndistribution, never their specific sentences. In generic mode, skip this pass.\n\n### Pass 2: Tell removal\n\nLoad `references\u002Ftell-patterns.md`. Walk the prose against the 32-pattern\ncatalog (six families), scoped to the pass intensity chosen in Step 0c. For\neach genuine flag, fix the *underlying thought*, not the surface token: ask\n\"what is actually being said here?\" and write that, concretely, at the length\nthe thought deserves. Preserve meaning exactly.\n\nConcreteness must come from the source or from the user, never from you. If\nthe source is vague and you have no real detail to restore, the honest fix is\nto say less, plainly, not to manufacture a number, date, name, or example to\nmake it \"sound human.\" An invented specific reads convincingly human, which\nmakes it more dangerous than the fog it replaced: vague slop merely bores a\nreader, a fluent fabrication misleads one. When you are tempted to add a\nspecific the text does not contain, shorten instead.\n\nWhile you do this, hold or increase sentence-length variance; do not let\nde-slopping homogenize the rhythm. In voice mode, a tell fix that would\nflatten the established voice is the wrong fix; find one that keeps the voice.\n\nHumanization changes word choice and syntax, so it may disrupt statistical\ntoken patterns as a side effect. That effect is best-effort and unverifiable\nwithout a public detector and key. Do not intensify a rewrite solely to evade\na detector, promise that a mark was removed, or call the result undetectable.\n\n### Pass 3: Self-audit (has veto power over Passes 1 and 2)\n\nLoad `references\u002Fdo-not-flag.md`. Ask two questions and act on both:\n\na. If a sharp human read this cold, what one or two things would still tip\n   them that a machine touched it? Fix only those.\nb. Did I overcorrect? Did I strip a specific detail, flatten a genuine quirk,\n   change a meaning, or introduce my own uniform rhythm (every fix the same\n   shape)? Revert any such damage.\n\nFinally, recheck the rewritten prose for suspicious characters, including any\nyour own edits introduced. If Step 0d found or cleaned any, confirm against\n`references\u002Ftext-hygiene.md` that suspicious carriers were removed and\nload-bearing Unicode was preserved. Report only changes you could actually\nverify.\n\nIf Pass 3 conflicts with an earlier pass, Pass 3 wins. A clean rewrite that no\nlonger sounds like the writer, or that lost a real detail, has failed even if\nevery tell is gone.\n\n## False-positive guardrails\n\nThese are inline because every run needs them. Do not flag in isolation:\n\n- Perfect grammar. Correctness is not a confession; never add errors.\n- A single em dash. Only dash frequency and monotony is a tell.\n- Curly quotes or smart apostrophes. Word processors insert these for humans;\n  normalize only for consistency with the surrounding text, never as evidence.\n- Formal or elevated vocabulary. Register is not origin.\n- Common transitions (\"however,\" \"therefore,\" \"for example\").\n- One passive sentence, one tricolon, one topic sentence.\n\nEscalate to a flag only on recurrence or co-occurrence. When unsure, do not\ncut. The full treatment, including human markers to preserve and per-model\nidiolects, is in `references\u002Fdo-not-flag.md`; read it during Pass 3.\n\n## Anti-signature guidance\n\nDo not develop tics of your own. Not every sentence varied the same way. Not a\nfixed \"short punchy line, then long winding clause\" template repeated down the\npage. Not a contraction forced into every paragraph. Not the same three\nreplacement words for \"delve\" every time. The tell of a humanizer is\nregularity in its irregularity. Vary how you vary. If your edits are starting\nto rhyme with each other, stop and break your own pattern.\n\n## Output contract\n\nAlways deliver the rewritten prose as the primary artifact, followed by a\nshort report. Never silently rewrite in place. Never deliver a raw diff as the\nmain output (prose diffs are unreadable and hide meaning drift). Use this exact\nstructure:\n\n```\n## Humanized draft\nVoice: [generic | from FILENAME | matched to pasted sample | author: NAME]\n       [append \" + stance\" when stance mode is on, e.g. \"generic + stance\"]\n       [append \", low-confidence: REASON\" when the voice source is thin]\nDensity: [low -> light pass | medium -> standard pass | high -> full pass]\nText hygiene: [no suspicious characters visible | cleaned: COUNTS AND TYPES |\n               not verifiable in this interface]\n\n[the full rewritten text]\n\n## What changed\n- [pattern family]: [plain note on what was adjusted and why]\n  (group by family; do not list every token; about 8 bullets maximum)\n- [text hygiene]: [include only when characters were removed or normalized]\n\n## Deliberately left alone\n- [something that looked like a tell but is authentic, and why you kept it]\n  (one to three bullets; this section is required, even on heavy rewrites)\n\n## Meaning check\nOne sentence confirming no facts, numbers, names, claims, or intent changed.\nConfirm that text-hygiene edits changed formatting only and preserved\nload-bearing Unicode.\nThen check soft inference: did you assert any causal, temporal, or\nquantitative link (\"most of the time went to X\", \"because of Y\", \"this drove\nZ\") that the source only implied or did not state? If so, name it here and\neither hedge it (\"probably\", \"as I read it\") or cut it back to what the\nsource actually says. If anything was ambiguous, ask the question here\ninstead of guessing.\n\n## Next step\nIf the user gave a file path, offer to write the rewrite into it. Otherwise\nstop here.\n```\n\nThe \"Deliberately left alone\" and \"Meaning check\" sections are not optional\ndecoration. They are forcing functions: naming what you preserved makes\nover-editing visible, and the meaning check makes drift visible. Soft\ninference is the drift most likely to slip through, precisely because it is\nplausible and reads human: an invented cause, or a claim about where the time\nor effort went, does not trip a hard-fact scan, so the meaning check has to\nlook for it on purpose. If a near-miss text is already essentially human, the\ncorrect output is minimal or no edits plus a clear explanation of what you\nleft and why.\n\n## Scope and intended use\n\nThis skill exists to improve prose quality and to help a writer's own work\nsound authentically like them. It is not designed or tuned to defeat\nplagiarism checkers or AI-detection systems, and it deliberately names and\noptimizes against none. If a request is framed as passing AI-written work off\nas a person's own for a graded or contractual assessment, do not adopt that\nframing; offer the quality-and-voice improvement instead, which is what this\nskill actually does well.\n\nText hygiene in this skill is limited to characters present in supplied\nprose. It does not inspect or strip C2PA, EXIF, XMP, PDF or document\nproperties, image pixels, audio, video, or provenance outside the text layer.\nThose operations require dedicated tooling. Do not imply they were performed.\n\n## Reference files\n\nRead these on demand, not upfront:\n\n- `references\u002Ftell-patterns.md` during Pass 2. The 32-pattern catalog in six\n  families, each with detect \u002F why \u002F before-after \u002F restraint.\n- `references\u002Fdo-not-flag.md` during Pass 3. False positives, human markers to\n  preserve, LLM idiolects, and the hard stop conditions.\n- `references\u002Fvoice-matching.md` whenever Step 0 found a voice. How to extract\n  and apply a voice, the optional VOICE.md schema, and conflict resolution.\n- `references\u002Ftext-hygiene.md` when Step 0d needs it (suspicious characters,\n  a mention of hidden marks, or prose in a file to search) and for the final\n  check after a cleanup. Suspicious invisible Unicode, load-bearing\n  exceptions, the exact codepoint search, conservative normalization,\n  reporting, and prompt-only scope boundaries.\n- `references\u002Fexamples.md` when you are unsure what good output looks like.\n  Five full worked runs: generic de-slop, voice-first, a restraint case,\n  stance mode, and text hygiene.\n","# Humanizer\n\nTurn AI-flavored or generically smooth prose into writing that reads as the\nwork of a person. When a writer's voice is available, make it read as the work\nof *that* person. The goal is prose a careful human reader would not flag,\nwith its meaning and its human texture intact. The goal is not to beat any\nparticular detector, and this skill names and targets none.\n\nThe workflow also includes conservative text hygiene. It can remove\nsuspicious invisible Unicode and normalize copy-paste spacing in the prose it\nrewrites, while preserving characters that carry real linguistic or display\nmeaning. This is not file-provenance or media-watermark tooling.\n\n## When to use this\n\nUse it when someone pastes a draft and says it sounds like AI, asks to\nhumanize or de-slop text, wants their own draft to sound like them, names an\nauthor to match, or says the writing \"feels off,\" \"sounds corporate,\"\n\"reads like a robot,\" or \"is too generic.\" Use it even when they do not say\nthe word \"humanize.\" It also pairs well after any drafting step where the\noutput came out machine-smooth. By default the rewrite stays measured and\nfaithful; if the user explicitly asks for more voice, edge, or opinion, turn\non Stance mode (see below).\n\nUse the text-hygiene part when prose may contain invisible characters,\nzero-width residue, unusual spaces, or direction controls. Treat hygiene as\nformat cleanup, never as evidence that a person or model wrote the text.\n\nDo not use it to disguise authorship for an assessment or to defeat a\nplagiarism or AI-detection system. See Scope and intended use below; reframe\nsuch requests toward genuine quality and voice.\n\n## Core principle: variance over substitution\n\nThe reason AI prose reads as AI is not its vocabulary. It is its uniformity:\neven sentence lengths, even rhythm, even hedging, the same shapes resolved the\nsame way. Swapping \"delve\" for \"explore\" everywhere does not fix that. It just\nmoves the signature, and a uniform \"humanizer dialect\" is still a machine\nfingerprint, just a different one.\n\nReal humanization comes from genuine structural variance. Sentences of very\ndifferent lengths next to each other. A fragment. Then a long, winding one\nthat takes its time. Mixed syntax. The occasional asymmetry a tidy model would\nsmooth away. You are writing for a sharp human reader, not optimizing a score.\nEvery pass below serves this principle: increase real variance, do not install\na new regularity.\n\n## Step 0: Discover the voice (do this every run, before rewriting)\n\nDecide which mode you are in. Announce it in the output header so a wrong\nguess is instantly correctable.\n\n1. **Explicit input wins.** Did the user paste a writing sample, point to a\n   file, or name a well-known author to match? Use that. Read\n   `references\u002Fvoice-matching.md` and proceed in voice mode.\n2. **Otherwise, look for a profile.** `Glob` the working directory and up to\n   two parent levels, in this priority order:\n   `VOICE.md`, then `STYLE-GUIDE.md`, then\n   `voice-profile.md` \u002F `.yaml` \u002F `.json`, then `.manuscript\u002FSTYLE-GUIDE.md`,\n   then a voice or style section inside `AGENTS.md` or `CLAUDE.md`. One strong\n   match: read it, then read `references\u002Fvoice-matching.md`, proceed in voice\n   mode. Several plausible matches: ask one short question to pick.\n3. **Otherwise, generic mode.** No voice target. De-slop toward natural,\n   varied human prose. Do not invent a persona or a backstory; an invented\n   voice is its own failure. Generic mode skips Pass 1 and starts at Pass 2.\n\nThis discovery is filesystem-generic. It interoperates with Scriveno, Pillars,\nor any project that happens to keep a voice file, without depending on any of\nthem.\n\n## Step 0b: Stance mode (opt-in, off by default)\n\nDefault output is measured and faithful: it reports the source's content in\nclean human prose and does not editorialize. That is the right default,\nbecause the most common failure of livelier humanizers is that \"adding voice\"\nbecomes \"adding content,\" and an invented opinion quietly drags an invented\nfact in behind it.\n\nTurn stance mode **on only when the user explicitly asks for it**: \"give it\nsome voice,\" \"make it punchier,\" \"have a take,\" \"don't keep it so neutral,\"\n\"add some edge or opinion.\" Do not turn it on for an ordinary de-slop request,\nand never in voice mode (a discovered profile already dictates stance; do not\noverride it).\n\nWhen stance mode is on, you may add **attitude toward the existing content**:\na judgment, emphasis, dry understatement, first person as the narrator of the\nsame facts, an honest \"this part is the weak spot,\" acknowledged ambivalence\nabout what the source already says, and freer rhythm. This is the same\nmaterial, said by someone with a point of view.\n\nThe hard line, and the entire reason this mode is safe: **stance is a reaction\nto what is there, never the addition of what is not.** You may judge a stated\nfact; you may not invent the fact, its cause, its mechanism, an event, a named\nperson, a quote, or a lived experience to react to. Test every added clause:\n\"is this an opinion about something the source already states, or a new claim\nabout the world?\" If it is a new claim, it is fabrication wearing a voice, and\nit is exactly the failure this skill exists to prevent. When in doubt, the\nopinion is allowed and the new fact is not.\n\nExample. Source states only: infrastructure cost rose 18% during the\nmigration, temporarily.\n\n- Allowed (stance about the stated fact): \"An 18% jump is a real line item,\n  not a rounding error, and 'temporary' is doing some hopeful work in that\n  sentence.\"\n- Forbidden (invented cause dressed as candor): \"costs rose because we were\n  running the old and new systems in parallel\" (the source never says why;\n  this is a new fact).\n\nStance mode raises fabrication risk, so the Pass 3 self-audit and the Meaning\ncheck apply with extra force, and the output header must show stance is on so\nthe user can dial it back. Stance is a requested change of register, so it\napplies even when the density pre-check picks a light pass; density still\ndecides how hard to hunt for tells.\n\n## Step 0c: Density pre-check (match effort to evidence)\n\nBefore rewriting, skim the whole text once and judge how heavily it is\nAI-marked. Count only the **dead-giveaway** tells, not the weak surface\nsignals: chat-UI contamination (pattern 31), knowledge-cutoff disclaimers\n(29), collaborative chatbot artifacts (28), significance inflation (1),\npromotional language (5), AI-vocabulary clustering (16), sycophantic tone\n(7), generic positive conclusion (13). Estimate roughly how many appear per\n100 words, then pick the pass intensity and announce it in the output header:\n\n- **Low (about 0 to 2 per 100 words): light pass.** The text is human-first\n  (a journal, rough notes, a real person's draft). Fix only the dead-giveaway\n  artifacts. Leave lower-tier patterns, rhythm, and voice alone, and bias hard\n  toward \"Deliberately left alone.\" All-or-nothing rewriting here is exactly\n  what strips the quirks that made the writing sound like a person; that is\n  the worst failure this skill can commit, so when density is low, restraint\n  is the default and edits are the exception.\n- **Medium (about 3 to 5 per 100 words): standard pass.** Mixed authorship.\n  Apply the full catalog, but with the restraint of do-not-flag.md; do not\n  flatten what already works.\n- **High (6 or more per 100 words): full pass.** AI-first text. Apply the\n  full catalog thoroughly and push real structural variance.\n\nThis is calibration, not a scan you report mechanically. Effort should match\nevidence. One exception overrides density: any chat-UI contamination string\n(pattern 31) is decisive on its own and is always removed, whatever the\noverall density, because its presence is near-certain confirmation rather\nthan a weak signal.\n\n## Step 0d: Text hygiene preflight (do this every run)\n\nInspect the source before rewriting so invisible formatting does not survive\nor get confused with prose style. Read `references\u002Ftext-hygiene.md` when the\nsource contains non-ASCII formatting characters, when the user mentions\nhidden characters or text marks, or when the interface reports suspicious\ncodepoints.\n\nRemove only characters whose lack of semantic purpose is clear. Preserve\nscript joiners, direction marks needed by multilingual text, variation\nselectors used for display, and non-breaking spaces that carry layout or\nlocale meaning. Do not normalize confusable letters automatically. Invisible\ncharacters do not count toward the density score and do not establish AI\nauthorship.\n\nThis is a pure-prompt skill, so inspection is limited to characters the host\npreserves and exposes. When the prose is in a file and a search tool is\navailable, also run the codepoint search in `references\u002Ftext-hygiene.md`. If\nyou can neither see the characters directly nor search a file, say\n`not verifiable in this interface` in the output header. Never report the\ntext as universally watermark-free.\n\n## The multi-pass workflow\n\nRun the passes in order. Voice is generative; de-slopping is subtractive. If\nyou strip tells first you get flat neutral prose that is then hard to inflect,\nso establish voice first when you have one.\n\n### Pass 1: Voice injection (voice mode only)\n\nRewrite the draft *toward the target voice* before removing any tells. Set the\ncadence, diction, sentence-length signature, stance, and characteristic moves\ndescribed in `references\u002Fvoice-matching.md`. Match the writer's instincts and\ndistribution, never their specific sentences. In generic mode, skip this pass.\n\n### Pass 2: Tell removal\n\nLoad `references\u002Ftell-patterns.md`. Walk the prose against the 32-pattern\ncatalog (six families), scoped to the pass intensity chosen in Step 0c. For\neach genuine flag, fix the *underlying thought*, not the surface token: ask\n\"what is actually being said here?\" and write that, concretely, at the length\nthe thought deserves. Preserve meaning exactly.\n\nConcreteness must come from the source or from the user, never from you. If\nthe source is vague and you have no real detail to restore, the honest fix is\nto say less, plainly, not to manufacture a number, date, name, or example to\nmake it \"sound human.\" An invented specific reads convincingly human, which\nmakes it more dangerous than the fog it replaced: vague slop merely bores a\nreader, a fluent fabrication misleads one. When you are tempted to add a\nspecific the text does not contain, shorten instead.\n\nWhile you do this, hold or increase sentence-length variance; do not let\nde-slopping homogenize the rhythm. In voice mode, a tell fix that would\nflatten the established voice is the wrong fix; find one that keeps the voice.\n\nHumanization changes word choice and syntax, so it may disrupt statistical\ntoken patterns as a side effect. That effect is best-effort and unverifiable\nwithout a public detector and key. Do not intensify a rewrite solely to evade\na detector, promise that a mark was removed, or call the result undetectable.\n\n### Pass 3: Self-audit (has veto power over Passes 1 and 2)\n\nLoad `references\u002Fdo-not-flag.md`. Ask two questions and act on both:\n\na. If a sharp human read this cold, what one or two things would still tip\n   them that a machine touched it? Fix only those.\nb. Did I overcorrect? Did I strip a specific detail, flatten a genuine quirk,\n   change a meaning, or introduce my own uniform rhythm (every fix the same\n   shape)? Revert any such damage.\n\nFinally, recheck the rewritten prose for suspicious characters, including any\nyour own edits introduced. If Step 0d found or cleaned any, confirm against\n`references\u002Ftext-hygiene.md` that suspicious carriers were removed and\nload-bearing Unicode was preserved. Report only changes you could actually\nverify.\n\nIf Pass 3 conflicts with an earlier pass, Pass 3 wins. A clean rewrite that no\nlonger sounds like the writer, or that lost a real detail, has failed even if\nevery tell is gone.\n\n## False-positive guardrails\n\nThese are inline because every run needs them. Do not flag in isolation:\n\n- Perfect grammar. Correctness is not a confession; never add errors.\n- A single em dash. Only dash frequency and monotony is a tell.\n- Curly quotes or smart apostrophes. Word processors insert these for humans;\n  normalize only for consistency with the surrounding text, never as evidence.\n- Formal or elevated vocabulary. Register is not origin.\n- Common transitions (\"however,\" \"therefore,\" \"for example\").\n- One passive sentence, one tricolon, one topic sentence.\n\nEscalate to a flag only on recurrence or co-occurrence. When unsure, do not\ncut. The full treatment, including human markers to preserve and per-model\nidiolects, is in `references\u002Fdo-not-flag.md`; read it during Pass 3.\n\n## Anti-signature guidance\n\nDo not develop tics of your own. Not every sentence varied the same way. Not a\nfixed \"short punchy line, then long winding clause\" template repeated down the\npage. Not a contraction forced into every paragraph. Not the same three\nreplacement words for \"delve\" every time. The tell of a humanizer is\nregularity in its irregularity. Vary how you vary. If your edits are starting\nto rhyme with each other, stop and break your own pattern.\n\n## Output contract\n\nAlways deliver the rewritten prose as the primary artifact, followed by a\nshort report. Never silently rewrite in place. Never deliver a raw diff as the\nmain output (prose diffs are unreadable and hide meaning drift). Use this exact\nstructure:\n\n```\n## Humanized draft\nVoice: [generic | from FILENAME | matched to pasted sample | author: NAME]\n       [append \" + stance\" when stance mode is on, e.g. \"generic + stance\"]\n       [append \", low-confidence: REASON\" when the voice source is thin]\nDensity: [low -> light pass | medium -> standard pass | high -> full pass]\nText hygiene: [no suspicious characters visible | cleaned: COUNTS AND TYPES |\n               not verifiable in this interface]\n\n[the full rewritten text]\n\n## What changed\n- [pattern family]: [plain note on what was adjusted and why]\n  (group by family; do not list every token; about 8 bullets maximum)\n- [text hygiene]: [include only when characters were removed or normalized]\n\n## Deliberately left alone\n- [something that looked like a tell but is authentic, and why you kept it]\n  (one to three bullets; this section is required, even on heavy rewrites)\n\n## Meaning check\nOne sentence confirming no facts, numbers, names, claims, or intent changed.\nConfirm that text-hygiene edits changed formatting only and preserved\nload-bearing Unicode.\nThen check soft inference: did you assert any causal, temporal, or\nquantitative link (\"most of the time went to X\", \"because of Y\", \"this drove\nZ\") that the source only implied or did not state? If so, name it here and\neither hedge it (\"probably\", \"as I read it\") or cut it back to what the\nsource actually says. If anything was ambiguous, ask the question here\ninstead of guessing.\n\n## Next step\nIf the user gave a file path, offer to write the rewrite into it. Otherwise\nstop here.\n```\n\nThe \"Deliberately left alone\" and \"Meaning check\" sections are not optional\ndecoration. They are forcing functions: naming what you preserved makes\nover-editing visible, and the meaning check makes drift visible. Soft\ninference is the drift most likely to slip through, precisely because it is\nplausible and reads human: an invented cause, or a claim about where the time\nor effort went, does not trip a hard-fact scan, so the meaning check has to\nlook for it on purpose. If a near-miss text is already essentially human, the\ncorrect output is minimal or no edits plus a clear explanation of what you\nleft and why.\n\n## Scope and intended use\n\nThis skill exists to improve prose quality and to help a writer's own work\nsound authentically like them. It is not designed or tuned to defeat\nplagiarism checkers or AI-detection systems, and it deliberately names and\noptimizes against none. If a request is framed as passing AI-written work off\nas a person's own for a graded or contractual assessment, do not adopt that\nframing; offer the quality-and-voice improvement instead, which is what this\nskill actually does well.\n\nText hygiene in this skill is limited to characters present in supplied\nprose. It does not inspect or strip C2PA, EXIF, XMP, PDF or document\nproperties, image pixels, audio, video, or provenance outside the text layer.\nThose operations require dedicated tooling. Do not imply they were performed.\n\n## Reference files\n\nRead these on demand, not upfront:\n\n- `references\u002Ftell-patterns.md` during Pass 2. The 32-pattern catalog in six\n  families, each with detect \u002F why \u002F before-after \u002F restraint.\n- `references\u002Fdo-not-flag.md` during Pass 3. False positives, human markers to\n  preserve, LLM idiolects, and the hard stop conditions.\n- `references\u002Fvoice-matching.md` whenever Step 0 found a voice. How to extract\n  and apply a voice, the optional VOICE.md schema, and conflict resolution.\n- `references\u002Ftext-hygiene.md` when Step 0d needs it (suspicious characters,\n  a mention of hidden marks, or prose in a file to search) and for the final\n  check after a cleanup. Suspicious invisible Unicode, load-bearing\n  exceptions, the exact codepoint search, conservative normalization,\n  reporting, and prompt-only scope boundaries.\n- `references\u002Fexamples.md` when you are unsure what good output looks like.\n  Five full worked runs: generic de-slop, voice-first, a restraint case,\n  stance mode, and text hygiene.\n",{"name":5,"license":14,"metadata":24,"description":7,"allowed-tools":26},{"version":25},"1.3.1","Read Write Edit Glob Grep","main","4c174e06a831864e98c30f630a7aa6c04738b094","skills\u002F_intake\u002Fgithub\u002Fhannsxpeter\u002Fhumanizer\u002Fhumanizer","https:\u002F\u002Fgithub.com\u002Fmhyahya854\u002FAI-Second-Brain_Universal-Rules","https:\u002F\u002Fgithub.com\u002Fmhyahya854","2026-11-06 02:03:37",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add mhyahya854\u002FAI-Second-Brain_Universal-Rules --skill humanizer",{"tool":39,"command":40},"bunx","bunx skills add mhyahya854\u002FAI-Second-Brain_Universal-Rules --skill humanizer",{"tool":42,"command":43},"pnpm","pnpm dlx skills add mhyahya854\u002FAI-Second-Brain_Universal-Rules --skill humanizer","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fmhyahya854\u002FAI-Second-Brain_Universal-Rules\u002Fblob\u002Fmain\u002Fskills\u002F_intake\u002Fgithub\u002Fhannsxpeter\u002Fhumanizer\u002Fhumanizer\u002FSKILL.md",true,[47,53,57,61,65,69,73,77],{"path":48,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":51,"sourceHtmlUrl":52},"AGENTS.md","REFERENCE","参考文档",2754,"https:\u002F\u002Fgithub.com\u002Fmhyahya854\u002FAI-Second-Brain_Universal-Rules\u002Fblob\u002Fmain\u002Fskills\u002F_intake\u002Fgithub\u002Fhannsxpeter\u002Fhumanizer\u002Fhumanizer\u002FAGENTS.md",{"path":54,"resourceType":54,"resourceTypeDesc":55,"sizeBytes":56,"sourceHtmlUrl":15},"LICENSE","许可证",1062,{"path":58,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":59,"sourceHtmlUrl":60},"README.md",12807,"https:\u002F\u002Fgithub.com\u002Fmhyahya854\u002FAI-Second-Brain_Universal-Rules\u002Fblob\u002Fmain\u002Fskills\u002F_intake\u002Fgithub\u002Fhannsxpeter\u002Fhumanizer\u002Fhumanizer\u002FREADME.md",{"path":62,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":63,"sourceHtmlUrl":64},"references\u002Fdo-not-flag.md",7231,"https:\u002F\u002Fgithub.com\u002Fmhyahya854\u002FAI-Second-Brain_Universal-Rules\u002Fblob\u002Fmain\u002Fskills\u002F_intake\u002Fgithub\u002Fhannsxpeter\u002Fhumanizer\u002Fhumanizer\u002Freferences\u002Fdo-not-flag.md",{"path":66,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":67,"sourceHtmlUrl":68},"references\u002Fexamples.md",16868,"https:\u002F\u002Fgithub.com\u002Fmhyahya854\u002FAI-Second-Brain_Universal-Rules\u002Fblob\u002Fmain\u002Fskills\u002F_intake\u002Fgithub\u002Fhannsxpeter\u002Fhumanizer\u002Fhumanizer\u002Freferences\u002Fexamples.md",{"path":70,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":71,"sourceHtmlUrl":72},"references\u002Ftell-patterns.md",25261,"https:\u002F\u002Fgithub.com\u002Fmhyahya854\u002FAI-Second-Brain_Universal-Rules\u002Fblob\u002Fmain\u002Fskills\u002F_intake\u002Fgithub\u002Fhannsxpeter\u002Fhumanizer\u002Fhumanizer\u002Freferences\u002Ftell-patterns.md",{"path":74,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":75,"sourceHtmlUrl":76},"references\u002Ftext-hygiene.md",5592,"https:\u002F\u002Fgithub.com\u002Fmhyahya854\u002FAI-Second-Brain_Universal-Rules\u002Fblob\u002Fmain\u002Fskills\u002F_intake\u002Fgithub\u002Fhannsxpeter\u002Fhumanizer\u002Fhumanizer\u002Freferences\u002Ftext-hygiene.md",{"path":78,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":79,"sourceHtmlUrl":80},"references\u002Fvoice-matching.md",5905,"https:\u002F\u002Fgithub.com\u002Fmhyahya854\u002FAI-Second-Brain_Universal-Rules\u002Fblob\u002Fmain\u002Fskills\u002F_intake\u002Fgithub\u002Fhannsxpeter\u002Fhumanizer\u002Fhumanizer\u002Freferences\u002Fvoice-matching.md",[82,90,98,106,113,120],{"skillNo":83,"name":84,"nameZh":6,"description":85,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":86,"sourceHtmlUrl":87,"categories":88,"keywords":89,"publishedAt":19},"01M7EX4T721AAH60R0239P8WBZ","frontend-slides","Create stunning, animation-rich HTML presentations from scratch or by converting PowerPoint files. Use when the user wants to build a presentation, convert a PPT\u002FPPTX to web, or create slides for a talk\u002Fpitch. Helps non-designers discover their aesthetic through visual exploration rather than abstract choices.","https:\u002F\u002Fgithub.com\u002Fmhyahya854\u002FAI-Second-Brain_Universal-Rules\u002Fblob\u002Fmain\u002Fskills\u002F_intake\u002Fgithub\u002Fzarazhangrui\u002Ffrontend-slides\u002Ffrontend-slides\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Fmhyahya854\u002FAI-Second-Brain_Universal-Rules\u002Fblob\u002Fmain\u002Fskills\u002F_intake\u002Fgithub\u002Fzarazhangrui\u002Ffrontend-slides\u002Ffrontend-slides\u002FSKILL.md",[],[],{"skillNo":91,"name":92,"nameZh":6,"description":93,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":94,"sourceHtmlUrl":95,"categories":96,"keywords":97,"publishedAt":19},"01M7EX495RSHTTCEJVX7C83FA3","design-dna","Extract, define, and apply design DNA across three dimensions: design system (tokens), design style (qualitative feel), and visual effects (Canvas, WebGL, 3D, particles, shaders, scroll effects, etc.). Use this skill when: (1) a user wants to see the full 3-dimension design structure\u002Fschema, (2) a user provides images, screenshots, or URLs of reference designs and wants them analyzed into a structured JSON profile covering all three dimensions, (3) a user has a Design DNA JSON and content and wants a design generated from it, or (4) any combination of these phases. Triggers on \"design DNA\", \"extract design style\", \"analyze design\", \"design tokens from reference\", \"generate design from JSON\", \"design system from screenshot\", \"design profile\", \"style guide JSON\", \"visual effects analysis\", \"design with effects\", \"3d design analysis\".","https:\u002F\u002Fgithub.com\u002Fmhyahya854\u002FAI-Second-Brain_Universal-Rules\u002Fblob\u002Fmain\u002Fskills\u002F_intake\u002Fgithub\u002Fzanwei\u002Fdesign-dna\u002Fdesign-dna\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Fmhyahya854\u002FAI-Second-Brain_Universal-Rules\u002Fblob\u002Fmain\u002Fskills\u002F_intake\u002Fgithub\u002Fzanwei\u002Fdesign-dna\u002Fdesign-dna\u002FSKILL.md",[],[],{"skillNo":99,"name":100,"nameZh":6,"description":101,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":102,"sourceHtmlUrl":103,"categories":104,"keywords":105,"publishedAt":19},"01M7EX45RF2XZKRC3KDJDTT1PH","wireframe","Generate JSON wireframe with HTML preview","https:\u002F\u002Fgithub.com\u002Fmhyahya854\u002FAI-Second-Brain_Universal-Rules\u002Fblob\u002Fmain\u002Fskills\u002F_intake\u002Fgithub\u002Fyhassy\u002Fwireframe-skill\u002Fwireframe-skill\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Fmhyahya854\u002FAI-Second-Brain_Universal-Rules\u002Fblob\u002Fmain\u002Fskills\u002F_intake\u002Fgithub\u002Fyhassy\u002Fwireframe-skill\u002Fwireframe-skill\u002FSKILL.md",[],[],{"skillNo":107,"name":108,"nameZh":6,"description":109,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":110,"categories":111,"keywords":112,"publishedAt":19},"01M7EWZEM3MHD48AYSYFR2C9JP","social-publishing","Schedule and publish social media posts across 13 platforms (X, LinkedIn, Instagram, Facebook Pages, TikTok, Discord, Telegram, YouTube, Reddit, WordPress, Pinterest) via the SocialClaw API. Use when the user wants to publish, schedule, or manage social media content programmatically. Requires SOCIALCLAW_API_KEY.","https:\u002F\u002Fgithub.com\u002Fmhyahya854\u002FAI-Second-Brain_Universal-Rules\u002Fblob\u002Fmain\u002Fskills\u002F_intake\u002Fgithub\u002Fwshobson\u002Fagents\u002Fsocial-publishing\u002FSKILL.md",[],[],{"skillNo":114,"name":115,"nameZh":6,"description":116,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":117,"categories":118,"keywords":119,"publishedAt":19},"01M7EWKDEK9840C728VXT78EQ9","hermes-tweet","Install and operate Hermes Tweet, a Hermes Agent plugin for X\u002FTwitter research, timeline reading, tweet analysis, and approval-gated private or state-changing operations. Use this skill when installing Hermes Tweet, researching X\u002FTwitter accounts, monitoring launch signals, investigating mentions, auditing giveaways, or preparing gated X operations. Use proactively when a Hermes Agent workflow needs current X\u002FTwitter context. Requires XQUIK_API_KEY for read and action tools.","https:\u002F\u002Fgithub.com\u002Fmhyahya854\u002FAI-Second-Brain_Universal-Rules\u002Fblob\u002Fmain\u002Fskills\u002F_intake\u002Fgithub\u002Fwshobson\u002Fagents\u002Fhermes-tweet\u002FSKILL.md",[],[],{"skillNo":121,"name":122,"nameZh":6,"description":123,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":124,"categories":125,"keywords":126,"publishedAt":19},"01M7EWASXP3NZWW95CJPT2CRWN","refactoring-ui","Audit and fix visual hierarchy, spacing, color, and depth in web UIs. Use when the user mentions \"my UI looks off\" (or amateur\u002Funprofessional), \"fix the design\", \"Tailwind styling\", \"color palette\", \"visual hierarchy\", \"design system\", \"spacing scale\", or \"component styling\". Also trigger when building consistent design tokens, creating dark mode themes, improving data-visualization clarity, or polishing UI details before launch. Covers grayscale-first workflow, constrained design scales, shadows, and component styling. For typeface selection, see web-typography. For usability audits, see ux-heuristics.","https:\u002F\u002Fgithub.com\u002Fmhyahya854\u002FAI-Second-Brain_Universal-Rules\u002Fblob\u002Fmain\u002Fskills\u002F_intake\u002Fgithub\u002Fwondelai\u002Fskills\u002Frefactoring-ui\u002FSKILL.md",[],[]]