agentleFS
Sign inStart free
daryaKutodaryaKuto/claude-skills/skills/unslop/SKILL.md

unslop

Skill0 starsChanged 16 days agoMITView on GitHub

Use when writing or editing prose a person will read (docs, READMEs, reports, specs, plans, PR descriptions, commit bodies, emails, blog or social posts), when asked to "unslop", "humanize", "de-AI" or "make this sound human", or when text reads generic, puffed up or machine-written.

197 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use principle-never-block-on-the-human, greploop-apps and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • jaydubya818Skill

    principle-never-block-on-the-human

    Apply when safe reversible implementation details can proceed inside scope.

    Also in 45 projects that use this

  • jaydubya818Skill

    greploop-apps

    Run greploop through @greptile-apps when the normal trigger rejects a large PR. Use only with the companion greploop skill.

    Also in 5 projects that use this

  • ianclearySkill

    principle-explain-the-number

    Apply before you trust, report, or act on a number you measured: a speedup, a regression, a throughput, a latency, or an eval result. Find what limits it, and rule out that it measured something other than the work you think.

    Also in 30 projects that use this

  • bnemaSkill

    setup-pstack

    Configure which Pi models pstack uses per role. Detects available provider/model specs and writes pstack's user configuration. Use for /setup-pstack, "configure pstack models", or changing pstack model choices.

    Also in 26 projects that use this

  • ianclearySkill

    poteto-help

    Guides users through pstack setup, /poteto-mode, and picking the skill, playbook, or principle for a task. Type /poteto-help with a question.

    Also in 15 projects that use this

  • +7 more from jaydubya818

The file

We split this skill into the rules it gives an agent

6 rules. 0 are common, 6 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 192 edited versions we read keep these.

  • 6

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Write so a software engineer intern can act on it, and so nobody can tell a model drafted it.

    line 9
  2. Distinctive:

    Not for code, config or data files.

    line 15
  3. Distinctive:

    If the text is outward-facing prose (README intro, blog, email, social post, landing page), apply Voice.

    line 26
  4. Distinctive:

    Stop after step 5. One pass, not a loop.

    line 30
  5. Distinctive:

    Use only facts from the source text or ones you have checked.

    line 69
  6. Distinctive:

    Use "I" or "we" when it fits.

    line 72

“Common” is measured: kept by at least 6 in 10 of the 192 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Unslop

Overview

Write so a software engineer intern can act on it, and so nobody can tell a model drafted it. Cut the patterns below, keep the meaning, check once more.

When to use

  • Any text a person will read: docs, reports, specs, plans, PR descriptions, commit bodies, code comments, emails, posts.
  • Not for code, config or data files.
  • For a full audit of a published or high-stakes document, use the anti-ai-writing skill (the complete Wikipedia catalog) instead.

The pattern

  1. Name the reader: an intern who can code but does not know this project.
  2. Shape the text for its form, following "Writing and tone" in the global CLAUDE.md. Technical documents keep their labelled structure (findings, options, "How we know"). Do not melt them into flowing prose.
  3. Scan for every pattern in the quick reference. Fix each hit: keep the meaning, use the plain version.
  4. If the text is outward-facing prose (README intro, blog, email, social post, landing page), apply Voice. For anything else, skip Voice.
  5. Self-audit once: "What still makes this read as AI-generated?" Fix it. Then search the text for the em dash character (U+2014): the count must be 0. Stop after step 5. One pass, not a loop. Do not report the pass; deliver the text.

Quick reference

Content

  • Significance inflation (pivotal moment, testament to, evolving landscape): say what happened.
  • Name-dropping outlets without context: pick one, say what it said.
  • Trailing -ing clauses (", highlighting...", ", ensuring..."): delete, or make it a real sentence.
  • Promotional words (vibrant, seamless, robust, groundbreaking): a specific fact. "Onboarding takes 3 days" beats "seamless onboarding".
  • Vague attribution (experts say, research suggests): name the source or cut.
  • "Despite challenges, it continues to thrive": give the specific fact.

Language

  • AI vocabulary (additionally, crucial, delve, enhance, fostering, garner, interplay, intricate, showcase, tapestry, underscore): plain words.
  • Copula avoidance (serves as, stands as, boasts, features): "is", "has".
  • "It's not just X, it's Y": state the point.
  • Forced groups of three: use the real count.
  • Synonym cycling: pick one term and repeat it.
  • False ranges ("from X to Y" with no real scale): list the items.

Style

  • Em dashes: none. Period, comma, colon or parentheses; "to" for ranges.
  • Colons as mid-sentence crutches: fine before a list or example, not as a connector.
  • Bold on every term: bold only what the reader must not miss.
  • Title Case Headings: sentence case.
  • Decorative emoji and curly quotes: remove; straight quotes only.

Communication and filler

  • Chatbot phrases (I hope this helps, Certainly!, Let me know if): cut.
  • Sycophancy (Great question!) and signposting (Let's dive in): just start.
  • Knowledge-cutoff disclaimers: find the source or cut.
  • Filler (in order to, due to the fact that, it is important to note): "to", "because", or cut.
  • Hedge stacks (could potentially possibly): one word, or commit to a position.
  • Summary endings (the future looks bright, in conclusion): end on the last new fact or the decision.

Voice (outward-facing prose only)

  • Have a position. React to facts instead of listing pros and cons neutrally.
  • Vary rhythm. Short sentences. Then a longer one that takes its time.
  • Be specific: "the scheduler fired 40 times overnight" beats "this is concerning". Use only facts from the source text or ones you have checked. Never add a feature, number or detail the source does not state; when a fact is missing, keep the sentence general or ask.
  • Use "I" or "we" when it fits.
  • Admit mixed results: "fast but fragile" beats "impressive".

Common mistakes

MistakeFix
Turning a report's labelled findings into prose to "sound human"Voice is for outward-facing prose only. Technical docs keep their structure.
Swapping one AI word for another (crucial to vital)Use the plain word or a specific fact.
Removing tells but leaving outward prose voicelessApply Voice.
Ending with a recapEnd on the last new fact.
Inventing specifics to sound concrete (features, numbers the source never gave)Every specific must come from the source or a check.
Narrating the pass ("I removed six AI patterns")Deliver the text only.

Source

Adapted from poteto/noodle, .agents/skills/unslop/SKILL.md (MIT licence), commit 82d2921c52370f23f29086de81ccfb600939c037, https://github.com/poteto/noodle. The pattern list and Voice section are adapted from it. Dropped from the original: "convert inline-header lists to prose" and "let some mess in", which clash with structured technical documents. Added: the intern reader, the technical-versus-outward split, signposting and summary endings. Full catalog: https://en.wikipedia.org/wiki/Wikipedia:Signs_of_AI_writing

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free