agentleFS
Pesty-MarketingPesty-Marketing/agent-skills/ai-writing-tells/SKILL.md

ai-writing-tells

Skill0 starsChanged 28 days agoMITView on GitHub

Anti-AI-slop audit for prose. Use before shipping any copy that goes out under the Pesty name or a client's: blog posts, emails, LinkedIn posts, reports, landing pages, case studies. Also use when asked to de-AI, humanize, check a draft for AI tells or AI slop, or make writing sound less like ChatGPT. Not for structure or narrative frameworks.

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    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:

    Run the script on the built HTML too, not just the markdown, so component copy is covered.

    line 20
  2. Distinctive:

    Commit to the claim or drop it.

    line 41

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “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

AI writing tells

Pesty's blog reader is an owner who has been pitched by four agencies and is scanning for anything that smells automated. These patterns are how they clock it. They also signal weak thinking: B and D claim significance instead of earning it (Zinsser, On Writing Well ch. 3: "If it should be pointed out, point it out.").

A real example, from a draft article on Yelp and ChatGPT leads that read as "littered with the AI style":

"And the honest counterweight, from the same data: weekend leads converted slightly better than weekday leads — 33.0% versus 30.2% — despite average replies that were markedly slower. Speed matters. Speed is not the only thing that matters, and anyone selling you speed as the whole answer is overselling it."

Four tells in three sentences: pre-announced honesty, an em-dash appositive, the X-matters/X-is-not-the-only-thing reversal, and a rule-of-three closer that escalates to a moral.

Done when

the bundled checker (bash scripts/check.sh <file>, run from this skill's folder; needs only bash and grep) returns no hits you haven't consciously kept, em-dash count is at or under target, and you've read the draft once for categories C and D, which regex only partly catches (two-beat reversals, rule-of-three moral closers, one-word dramatic paragraphs).

Run the script on the built HTML too, not just the markdown, so component copy is covered.

The checklist

A. Announcing significance instead of showing it. "the distinction matters" · "that matters more than it sounds" · "X matters. X is not the only thing that matters." · "worth reading twice" · "worth being blunt" · "worth knowing about" · "it is worth noting" · "the part that matters is" · "and that is the point" · "this is the finding that changes the decision" · "here's the thing"

B. Pre-announcing your own honesty. "To be precise, because…" · "To be clear" · "the honest version is this" · "the honest counterweight" · "Honestly, it depends" · "we want to be careful here" · "we will not claim either way" · "let me be direct" · "full disclosure" · "read it that way round"

C. The antithesis tic. "It is not X. It is Y." · "not just X, but Y" · "X — and that is exactly the problem" · two-beat reversals · rule-of-three closers that build to a moral

D. False-profundity fragments. One-word dramatic paragraphs ("No." "Yes." "It is real.") · "Now put X next to it." · "Read that again." · "Hold onto that distinction." · "That is the whole story."

E. Filler intensifiers and hedge-stacking. genuinely · truly · meaningful/meaningfully · "some meaningful share of" · "it is entirely possible that" · arguably · "at the end of the day" · "make no mistake" · ("actually" is fine when it contrasts with something stated, not as an intensifier)

F. Vocabulary tells. delve · leverage (verb) · landscape · navigate · robust · underscore · crucial · pivotal · testament · seamless · game-changer · double-edged · nuanced · "in today's…" · "the reality is"

G. Punctuation habit. Em-dash appositives in consecutive sentences; the trailing "— and that is…" clause. Roughly one em-dash per 250 words reads human; five in a row does not.

Fixing a hit

  • Delete rather than reword. Zinsser's bracket test: if the sentence survives without the phrase, cut it. Swapping one tell for another is the common failure.
  • Commit to the claim or drop it. "The honest counterweight is X" becomes "X." If X isn't worth stating plainly, it wasn't worth a preamble.
  • Supporting reading: Zinsser, On Writing Well, ch. 2 "Simplicity", ch. 3 "Clutter" (bracket test), ch. 4 "Style" (don't hedge), ch. 9 "The Lead and the Ending" (stop when ready; no summary conclusions).
  • Blog/SEO copy also targets Flesch-Kincaid grade 11.

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.