agentleFS
ragnos-labsragnos-labs/skill-index/skills/ai-writing-review/SKILL.md

ai-writing-review

Skill0 starsChanged 3 months agoMITView on GitHub

Find, grade, and explain common AI-like writing patterns in supplied text and recommend repairs. Use for the user's own drafts or client writing; this is a prose-pattern review, not proof of authorship. Use writing-cleanup when the primary request is a rewrite.

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 3 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read references/pattern-catalog.md for the full catalog.

    line 26
  2. Distinctive:

    Do not average the scores into an "AI probability." Lead with the two or three families that most affect quality.

    line 52
  3. Distinctive:

    Do not add anecdotes, personal experiences, quotations, sources, or results to make the writing seem human.

    line 67
  4. Very specific:

    When cleaning is requested, apply the repair plan through writing-cleanup and return the revised text first, followed by a compact change …

    line 81

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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 Review

Identify writing patterns that often make generated or heavily templated prose feel generic. Improve the text without flattening it into another house style.

Choose the mode

  • Detect: mark passages and explain the patterns present.
  • Grade: score pattern families and identify the highest-impact repairs.
  • Compare: evaluate before and after versions for real improvement and lost information.
  • Review and clean: record the findings, then apply writing-cleanup when the user requests a revised artifact.

This skill cannot establish whether a person used AI from prose alone. Do not report an authorship probability or accuse a writer. Report observable language patterns, confidence in each finding, and the effect on the reader.

Inspect the text

Read references/pattern-catalog.md for the full catalog. Look for clusters rather than treating one phrase as proof.

Evaluate:

  • generic framing and unsupported significance;
  • predictable structure and repetitive section jobs;
  • sentence rhythm, variation, and over-regularity;
  • vague attribution, empty authority, or fabricated specificity;
  • unnecessary signposting, recaps, and conclusion repetition;
  • inflated, abstract, promotional, or synthetic-sounding diction;
  • excessive hedging, qualification, or balanced-sounding filler;
  • canned contrasts, triplets, fragments, and rhetorical questions;
  • publication residue such as prompts, notes, or meta-commentary; and
  • factual, citation, quotation, or meaning drift introduced by prior rewrites.

Grade

Score each applicable family from 0 to 4:

  • 0: absent or negligible;
  • 1: occasional and harmless;
  • 2: noticeable but localized;
  • 3: frequent and distracting; and
  • 4: dominant enough to make the piece feel generic or unreliable.

Do not average the scores into an "AI probability." Lead with the two or three families that most affect quality.

Recommend repairs

  1. Lock names, facts, quantities, dates, quotations, citations, links, and commitments.
  2. Repair structure before polishing sentences.
  3. Replace generic claims with supported specifics or remove them.
  4. Vary rhythm by following the thought, not by randomly changing sentence length.
  5. Restore the author's real directness, warmth, technical depth, humor, or rough edges when evidence of that voice exists.
  6. Compare the revision with the original for lost meaning or support.

Do not add anecdotes, personal experiences, quotations, sources, or results to make the writing seem human. Do not deliberately insert errors, slang, or awkwardness to evade a detector.

Deliver

For detection or grading, return:

  1. overall prose-quality judgment;
  2. highest-impact pattern findings with excerpts;
  3. dimensional scores;
  4. recommended repair order; and
  5. an authorship caveat only if the user asked whether AI wrote it.

When cleaning is requested, apply the repair plan through writing-cleanup and return the revised text first, followed by a compact change note and any factual or source questions that remain.

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.