agentleFS
abduallahJamalabduallahJamal/humanizer-skills/.agents/skills/academic-humanizer/SKILL.md

academic-humanizer

Skill0 starsChanged 10 days agoMITView on GitHub

Humanize formal, research, scientific, legal, and academic text in English or Modern Standard Arabic without introducing casual slang, contractions, or broken citations. Preserves citations, p-values, equations, and domain terms while disrupting AI rhythm and nominalization bloat (أنسنة الأبحاث العلمية والرسائل الأكاديمية والتقارير الرسمية بالعربية والإنجليزية).

10 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

10 rules. 0 are common, 1 is distinctive, and 9 are very specific to this project.

  • 0

    Common

    Most of the 10 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 9

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Zero Citation or Number Drift: Every inline citation ([1], [12–15], (Smith et al., 2025), (العلي، 2024)), DOI, equation ($...$, $$...$$), …

    line 18
  2. Very specific:

    Arabic (الفصحى الأكاديمية الرصينة): No colloquialisms, no journalistic clickbait, and no translationese (تم إجراء / يقوم بدراسة).

    line 21
  3. Very specific:

    Domain Terminology Shield: Never swap established technical, medical, legal, or statistical terms for thesaurus synonyms (homoscedasticity, …

    line 22
  4. Very specific:

    Read references/academic-transforms.md for position-aware syntactic inversions, denominalization rules, and scholarly Arabic …

    line 24
  5. Very specific:

    When python3 is available, run prep mode to expand contractions, strip copy-paste citation artifacts (oaicite, zero-width spaces), and …

    line 29
  6. Very specific:

    Ban stock transition spam: Do not prepend Moreover,, Furthermore,, Additionally,, علاوة على ذلك،, or بالإضافة إلى ذلك، to paragraphs.

    line 35
  7. Very specific:

    Instead of casual fragments, create sentence-length variance (8–34 words) using authentic academic syntax:

    line 40
  8. Very specific:

    Balanced active/passive voice: Use active voice for your methodology/claims (We normalize token counts...

    line 43
  9. Distinctive:

    Verify that zero citations, numbers, or fixed terms were lost, and optionally generate a word-level change diff:

    line 46
  10. Very specific:

    If verify exits non-zero, restore every missing citation or number before returning the text.

    line 50

“Common” is measured: kept by at least 6 in 10 of the 10 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

Academic & Formal Humanizer (EN & MSA Arabic)

General humanizers inject contractions (don't, it's), fragments (Like this.), and conversational asides (Honestly,) that break scholarly register. Conversely, naive rule-based academic rewriters prepend "Moreover," and "Furthermore," everywhere, which increases AI detection scores.

This skill humanizes research papers, literature reviews, theses, grant proposals, executive reports, and formal Arabic/English prose while keeping full scholarly rigor and 100% citation/statistic integrity.

Mandatory Integrity Locks

  1. Zero Citation or Number Drift: Every inline citation ([1], [12–15], (Smith et al., 2025), (العلي، 2024)), DOI, equation ($...$, $$...$$), sample size (N = 312), confidence interval, and p-value (p < 0.001) must survive verbatim.
  2. Formal Register Guard:
    • English: Expand all contractions (cannot, do not, does not, is not). No colloquialisms (stuff, kind of, turns out), no standalone dramatic fragments, and no first-person I think unless the paper's discipline explicitly uses first-person plural (we observe).
    • Arabic (الفصحى الأكاديمية الرصينة): No colloquialisms, no journalistic clickbait, and no translationese (تم إجراء / يقوم بدراسة). Use authentic scholarly Arabic verbs (أظهرت النتائج، فحصت الدراسة، يُستدل من ذلك، تتفق هذه النتيجة مع).
  3. Domain Terminology Shield: Never swap established technical, medical, legal, or statistical terms for thesaurus synonyms (homoscedasticity, gradient descent, الشريعة, الزكاة, الانحدار اللوجستي stay exact).

Read references/academic-transforms.md for position-aware syntactic inversions, denominalization rules, and scholarly Arabic transformations.

The 4-Step Scholarly Workflow

Step 1: Deterministic Pre-Processing (academic_prep.py)

When python3 is available, run prep mode to expand contractions, strip copy-paste citation artifacts (oaicite, zero-width spaces), and inventory all citations and numbers that must survive:

python3 <skill-dir>/scripts/academic_prep.py prep --input draft.txt --output prepped.txt

Step 2: Strip Academic AI Tells (Without Adding Stock Transitions!)

  • Ban stock transition spam: Do not prepend Moreover,, Furthermore,, Additionally,, علاوة على ذلك،, or بالإضافة إلى ذلك، to paragraphs. Connect sentences through referential cohesion (This discrepancy..., Such constraints..., ويعزى هذا التباين إلى..., غير أن هذا الافتراض يصطدم بـ...).
  • Cut empty significance inflation: Replace "plays a pivotal role in the evolving landscape of X" with the exact mechanism or empirical scope ("governs latency in distributed consensus").
  • Cut hollow literature-review boilerplate: Replace "Numerous studies have extensively investigated X" with direct synthesis of what the cited works actually found ("Prior work links X to Y [3, 7], yet leaves Z untested").

Step 3: Syntactic Inversion & Denominalization (Natural Scholarly Burstiness)

Instead of casual fragments, create sentence-length variance (8–34 words) using authentic academic syntax:

  1. Position-aware denominalization: Turn noun-heavy padding ("benefit from the implementation of quantization" / "تحقيق الاستفادة من عملية تطبيق التكميم") into crisp verbal/gerund clauses ("benefit from quantizing weights" / "الاستفادة من تكميم الأوزان"). Never leave dangling "implementing of".
  2. Clause-order inversion: Alternate subject-first declaratives (We evaluated three architectures across four benchmarks.), subordinate-first framing (When packet loss exceeds 2%, throughput collapses under Cubic.), and concise empirical anchors (Table 2 summarizes the ablation.).
  3. Balanced active/passive voice: Use active voice for your methodology/claims (We normalize token counts... / حلّلنا العينة...) and natural passive only where the process or specimen is the true topic (Samples were annealed at 400 °C / عُولجت العينات عند درجة...).

Step 4: Citation, Digit & Diff Verification

Verify that zero citations, numbers, or fixed terms were lost, and optionally generate a word-level change diff:

python3 <skill-dir>/scripts/academic_prep.py verify --original draft.txt --rewritten final.txt --diff diff.md

If verify exits non-zero, restore every missing citation or number before returning the text.

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.