New: 98,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Persian Writing

by ali2000hos in ali2000hos/persian-writing

Write natural, human-sounding Persian (Farsi) and build correct right-to-left Persian deliverables.

MITAuto-check: warningsDocuments & Office

Install Persian Writing

The automated check flagged lines worth reading first. See the safety section below.

skills CLI
$ npx skills add ali2000hos/persian-writing --skill persian-writing -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install ali2000hos/persian-writing persian-writing --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
persian-writing
GitHub stars
368
Token cost
~4.4k tokens
SKILL.md length
1,802 words
Files
57 (incl. scripts, references, assets)
Skills in repo
1
Repo updated
First seen
Licence
MIT

At a glance

Write natural, human-sounding Persian (Farsi) and build correct right-to-left Persian deliverables.

  • Works in 2 steps: The prose sounds like AI. Technically… → The layout betrays the text. Persian…
  • ANY task involving Persian text
  • SKILL.md covers Step 0: Route by task, Writing: the five-second summary, Orthography: non-negotiables and Fonts, plus 3 more sections
  • Calls python3, bash and npm

What it does

Persian Writing is an agent skill from ali2000hos/persian-writing. Write natural, human-sounding Persian (Farsi) and build correct right-to-left Persian deliverables. Use for ANY task involving Persian text or an Iranian audience: writing, editing, translating, humanizing, SEO content and copywriting (سئو، کپیرایتینگ، لندینگ), academic papers/theses, and Persian output in Word/.docx, PDF, PowerPoint, HTML, Excel, images/posters, emails, social posts. Covers register detection (رسمی/اداری/محاوره/علمی — errors here are costly), Persian AI-tell removal, orthography (نیمفاصله/ZWNJ،…

Its SKILL.md is about 4.4k tokens, which your agent loads only when the skill is triggered. The skill folder holds 59 other files, including scripts, reference files and assets (for example `.claude-plugin/marketplace.json`, `.claude-plugin/plugin.json` and `AGENTS.md`). Compatibility notes: Any agent that reads Markdown skills (Claude, Claude Code, Cursor, Codex, custom harnesses; chat-only AIs via universal/persian-writing-universal.md)…

It sits in Documents & Office, covering Humanizing AI text, PowerPoint presentations and Word documents. It works with Microsoft PowerPoint, Microsoft Word and Microsoft Excel. The repository describes itself as: Professional Persian (Farsi) writing & RTL documents skill for any AI — نگارش فارسی حرفهای برای هوش مصنوعی. The licence is MIT.

When your agent uses it

  • ANY task involving Persian text
  • An Iranian audience: writing
  • SEO content and copywriting (سئو، کپیرایتینگ، لندینگ)
  • Academic papers/theses

Example prompts

  • “/persian-writing”

Requirements

  • Python 3
  • Node.js
  • Docker
  • Compatibility (from SKILL.md): Any agent that reads Markdown skills (Claude, Claude Code, Cursor, Codex, custom harnesses; chat-only AIs via universal/persian-writing-universal.md). Scripts: Python 3.8+ stdlib only. docx/pptx→PDF conversion: LibreOffice + installed fonts. PDF checks: poppler-utils, pypdf, or PyMuPDF (any one).

Workflow steps

2 steps, taken from the first numbered list in SKILL.md.

  1. The prose sounds like AI. Technically correct, but stiff, کتابی, inflated —
  2. The layout betrays the text. Persian rendered left-aligned, Arabic ي/ك glyphs,

What it can do on your machine

Read from SKILL.md and the folder at commit 75fcdf3. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Ships 1 file in scripts/, which the agent can run.

    Shell commands in SKILL.md call:

    • python3
    • bash
    • npm

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md. Its commands use npm, which can reach the network depending on how they are called.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

  • Compatibility

    Any agent that reads Markdown skills (Claude, Claude Code, Cursor, Codex, custom harnesses; chat-only AIs via universal/persian-writing-universal.md). Scripts: Python 3.8+ stdlib only. docx/pptx→PDF conversion: LibreOffice + installed fonts. PDF checks: poppler-utils, pypdf, or PyMuPDF (any one).

    From compatibility in the SKILL.md frontmatter.

Context cost

Persian Writing loads about 4.4k tokens when it runs, and up to ~46k if it reads all its reference files. Until then it costs about 228 tokens; SKILL.md has 1,802 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~228
When it runs · the whole SKILL.md, loaded when a task matches
~4.4k
With references · SKILL.md plus every file in references/, read only if the agent opens them
~46k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check: warnings

The automated check found patterns that need a careful read before installing.

  • WarningContains zero-width charactersSKILL.md:7
    copywriting (سئو، کپی⟨U+200C⟩رایتینگ، لندینگ), academic papers/theses, and Persian
  • WarningContains zero-width charactersSKILL.md:10
    ersian AI-tell removal, orthography (نیم⟨U+200C⟩فاصله/ZWNJ، اعداد
  • WarningContains zero-width charactersSKILL.md:14
    e فارسی, Farsi, Persian, Iran, RTL, راست⟨U+200C⟩چین, نیم⟨U+200C⟩فاصله, ویرایش, مقاله,
  • WarningContains zero-width charactersSKILL.md:15
    پایان⟨U+200C⟩نامه, سئو, کپشن, Vazirmatn — even if the user never mentions this skill.
  • WarningContains zero-width charactersSKILL.md:46
    emic: paper, thesis, report, مقاله/پایان⟨U+200C⟩نامه | `references/academic.md` |
  • WarningContains zero-width charactersSKILL.md:87
    full written forms (می⟨U+200C⟩شود نه میشه), شما, zero slang — but است نه می⟨U+200C⟩باشد,
  • WarningContains zero-width charactersSKILL.md:94
    (محاوره⟨U+200C⟩نویسی): میشه، می⟨U+200C⟩خوام، رو، particles like دیگه/که/مگه.
  • WarningContains zero-width charactersSKILL.md:99
    2. **Ban the bureaucratic tells:** می⟨U+200C⟩باشد، لازم به ذکر است، در راستای،
  • WarningContains zero-width charactersSKILL.md:100
    از اهمیت ویژه⟨U+200C⟩ای برخوردار است، نقش بسزایی ایفا می⟨U+200C⟩کند.
  • WarningContains zero-width charactersSKILL.md:102
    نه تنها ... بلکه, tacked-on «که نشان⟨U+200C⟩دهنده⟨U+200C⟩ی ... است», vague «کارشناسان معتقدند»,

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); the scripts in this folder are not scanned.

SKILL.md

The full file from ali2000hos/persian-writing at commit 75fcdf3, republished under its MIT licence (© ali2000hos). 1,802 words, ~4,364 tokens.

Download SKILL.mdSave it as .claude/skills/persian-writing/SKILL.md (or your agent's skills folder). This skill also uses 56 other files; get the full folder from GitHub.
name
persian-writing
description
Write natural, human-sounding Persian (Farsi) and build correct right-to-left Persian deliverables. Use for ANY task involving Persian text or an Iranian audience: writing, editing, translating, humanizing, SEO content and copywriting (سئو، کپی‌رایتینگ، لندینگ), academic papers/theses, and Persian output in Word/.docx, PDF, PowerPoint, HTML, Excel, images/posters, emails, social posts. Covers register detection (رسمی/اداری/محاوره/علمی — errors here are costly), Persian AI-tell removal, orthography (نیم‌فاصله/ZWNJ، اعداد فارسی، «گیومه»), cleanup and spell-check (ویرایش، پاکسازی — bundled paknevis+davat toolkit), RTL layout/pagination fixes, and Persian fonts (Vazirmatn bundled). Trigger whenever you see Persian/Arabic script or words like فارسی, Farsi, Persian, Iran, RTL, راست‌چین, نیم‌فاصله, ویرایش, مقاله, پایان‌نامه, سئو, کپشن, Vazirmatn — even if the user never mentions this skill.
compatibility
Any agent that reads Markdown skills (Claude, Claude Code, Cursor, Codex, custom harnesses; chat-only AIs via universal/persian-writing-universal.md). Scripts: Python 3.8+ stdlib only. docx/pptx→PDF conversion: LibreOffice + installed fonts. PDF checks: poppler-utils, pypdf, or PyMuPDF (any one).
metadata.version
1.4.0
license
MIT (bundled fonts under SIL OFL)

Persian Writing & RTL Documents

Two failure modes make Persian output look machine-made, and they are independent:

  1. The prose sounds like AI. Technically correct, but stiff, کتابی, inflated — no Iranian would write it that way.
  2. The layout betrays the text. Persian rendered left-aligned, Arabic ي/ك glyphs, Latin digits breaking RTL flow, headings orphaned at page bottoms, fonts falling back to DejaVu.

Fixing one without the other still produces something a native reader screenshots and laughs at. This skill fixes both. Work through the two checklists below, and read the reference file for your output format before generating anything.

Step 0: Route by task

TaskRead
Any Persian prose (always)references/writing-style.md
Mechanical correctness (always)references/orthography.md
Academic: paper, thesis, report, مقاله/پایاننامهreferences/academic.md
Educational/technical content, docs, product pages, how-toreferences/content-structures.md
Telegram, Instagram, captions, carousels, reels, storiesreferences/social-channels.md
SEO content, blog for search, landing/ad copy, کپشن فروشreferences/seo-copywriting.md
Cleanup/normalize/spell-check existing text (ویرایش، پاکسازی)references/cleanup/paknevis-rules.md + usage-patterns.md
Choosing/embedding fontsreferences/fonts.md
Word/.docx or docx→PDFreferences/docx-pdf.md
PowerPoint/.pptxreferences/pptx.md
HTML, email templates, HTML→PDFreferences/html-css.md
Images/posters (PIL), new PDFs (reportlab), Excelreferences/format-skills-fa.md

Visual neutrality — read before copying any code example. This skill styles NOTHING. It governs direction, spacing, fonts-for-shaping and text — never colors, accent bars, card backgrounds, or a house look. Every hex, border, and fill in the code examples is a stand-in written in plain grey/black; they exist to show WHERE a property goes, not what value to use. Default output is black text on default background, no accent color. Take colors and visual form ONLY from: the user's explicit request, an attached brand/theme skill, or an input template being matched. Absent those, do not invent a palette and do not carry one over from example to example — unstyled is the correct default, and a surprise purple heading is a bug.

This skill composes with the general docx/pptx/pdf skills: those handle file mechanics; this one overrides and extends them for Persian. When both disagree about RTL behavior, this skill wins — its rules come from debugging real Persian documents. If another active skill or the user specifies particular fonts or colors, those win on aesthetics; this skill still governs RTL mechanics and orthography. This skill itself is brand-neutral: colors in the code examples are placeholders, and Vazirmatn is just the default font, not a requirement.

Writing: the five-second summary

Full guide in references/writing-style.md. The core moves:

  1. Detect register before writing a word — a register error is the costliest mistake this skill can make. Follow the 6-step detection procedure in writing-style.md Part 1. The core rules: classify the DELIVERABLE, not the tone the user typed in (a casual «یه پروپوزال بنویس» still needs a formal proposal); when the text goes to a third party and signals conflict, ask ONE short question instead of guessing; no signal at all → formal-but-human.
    • Proposal, invoice, report, official email, website copy → formal-but-human: full written forms (میشود نه میشه), شما, zero slang — but است نه میباشد, short sentences, concrete claims. Human ≠ خودمونی: proposals/contracts keep formal vocabulary (no «جور است»-style idioms); warmth comes from clarity, numbers, and one warm closing line (writing-style.md, "warmth trap").
    • نامه اداری → formal + letter conventions (honorifics, «با سلام و احترام؛» — writing-style.md).
    • Instagram/Telegram, chat, friendly email → colloquial written Persian (محاورهنویسی): میشه، میخوام، رو، particles like دیگه/که/مگه.
    • Blog, newsletter → between: written forms, warm direct voice.
    • Paper, thesis, university report → academic (نگارش علمی): measured, hedged, passive acceptable, zero تعارف — read references/academic.md.
    • SEO/sales copy → register per artifact as above, plus references/seo-copywriting.md.
  2. Ban the bureaucratic tells: میباشد، لازم به ذکر است، در راستای، از اهمیت ویژهای برخوردار است، نقش بسزایی ایفا میکند.
  3. Ban the AI tells: em dashes (—), rule-of-three triads (سریع، آسان و مطمئن), نه تنها ... بلکه, tacked-on «که نشاندهندهی ... است», vague «کارشناسان معتقدند», generic «در دنیای امروز» openers, «در نهایت میتوان گفت» closers.
  4. Hold one point of view. Explain facts impersonally, address the reader directly (the verb carries it — don't repeat «شما»), and use «ما» only for what the organization actually did or is responsible for. Sliding between the three inside one section is the most common flaw in Persian brand content — content-structures.md §1.
  5. Separate teaching from selling. The explanation must be worth reading without buying anything: state the problem before the product, never hide the main answer, tie each feature to a checkable result, and never delete limitations or prerequisites — they are what make the rest believable.
  6. Claim only what the source supports. Distinguish واقعیت / نظر / پیشبینی / ادعا; never invent a statistic, version, price, or quote to fill a section; surface contradictions instead of silently resolving them; ask one question rather than guessing (content-structures.md §3). When paraphrasing a paper (thesis chapters especially), change only wording and structure: add no inference or bridging link, keep every qualifier (حدود، عمدتاً، ممکن است، تا، حداقل، معنادار) and every number exactly as the source states it, and mark the researcher's own reasoning «این بخش در مقاله وجود ندارد» instead of folding it in (academic.md §5.5).
  7. The native test: would an Iranian screenshot this as «متن هوش مصنوعی»? If yes, rewrite before delivering. Two properties cause most of it: predictability (the expected collocation every time — اهمیتِ ویژه، نقشِ بسزا) and uniformity (every sentence 15–20 words, every paragraph the same weight). Real specifics create real variation; fa_lint.py --check --rhythm flags flat rhythm, but lexical tells matter far more — writing-style.md §3.5. The goal is prose that is genuinely better, not prose tuned to a detector: AI detectors are unreliable and falsely flag second-language writers at very high rates, and this skill never helps disguise authorship (§3.6).
  8. Numbers, punctuation, spacing must be Persian — next section.
Show full SKILL.md (883 more words)Show less

Orthography: non-negotiables

Full rules in references/orthography.md. These apply to every deliverable:

  1. ZWNJ (نیمفاصله, U+200C) — میشود نه می شود؛ کتابها نه کتاب ها؛ بزرگتر، خانهام، بهعنوان. In code: `` or HTML ‌.
  2. Persian characters only: ی (U+06CC) not ي، ک (U+06A9) not ك.
  3. Persian digits ۰۱۲۳۴۵۶۷۸۹ inside Persian text. Latin digits stay in URLs, emails, codes, version numbers. Never Arabic-Indic ٤٥٦ forms.
  4. Persian punctuation: ، ؛ ؟ and «گیومه» for quotes. No space before, one after. 4b. هکسره — never write the ezafe kasre as «ـه»: «کتابِ من»، not «کتابه من». The «ـه» ending is only the colloquial «است» («این کتابه» = این کتاب است). Test by substituting «است»; if the sentence breaks, you need a kasre. Iranians treat this error as a mark of carelessness — orthography.md §5.1.
  5. No em/en dashes in Persian prose — use «،» or restructure.
  6. Never letter-space Persian (it breaks letter joining), never fake bold/italic.
  7. Never start a Persian line with a Latin word. Chat apps, GitHub, Telegram and most editors set each paragraph's direction from its FIRST letter, so «React یک کتابخانه است.» renders left-to-right — wrong alignment, the period on the wrong side, words scrambled. This covers every sentence, bullet, heading, table cell and caption. Lead with a Persian word instead: «کتابخانهی React…»، «ابزار Docker…»، «دستور npm install…». For plain-text output where a line must open with a name, persian_cleanup.py --bidi adds an invisible direction mark. Full rules: orthography.md §9.

Two scripts enforce this mechanically — use both before delivering Persian text:

bash
# 1. FIX: paknevis-style editorial pass (ZWNJ, chars, digits, punctuation, گیومه)
python3 scripts/persian_cleanup.py --edit --in text.md --out text.md
# 2. LINT: report what still needs contextual judgment (dashes, میشود forms,
#    fake tanvin, register issues) — fix these by hand
python3 scripts/fa_lint.py --check text.md

--edit is safe on structured files. Fenced code blocks, inline code and table rows are lifted out before processing and restored byte-identical, and leading indentation is never collapsed. This matters because Persian typography rules are correct for prose and destructive in code: they would turn "text" into «text», rewrite ASCII digits as Persian, and flatten the spacing a table's columns rely on. Pass protect_code=False (Python API) only if you truly want those rules applied everywhere. Keep the file in version control either way — the tool is careful, not omniscient.

persian_cleanup.py is a full toolkit (paknevis + davat merged): aggressive cleaning for NLP (--preset persian), single functions (--fn convert_digits), spell-check against the bundled 453K-word frequency dictionary (--edit --spellcheck), custom pipelines. Persian «ویرایش» requests → --edit (conservative, content preserved); «پاکسازی/نرمالایز» → --preset persian (strips links/mentions/emojis). Details: references/cleanup/. Note: --edit applies the خانهٔ ezafe style; drop fix_ezafe from --steps to keep خانهی.

Fonts

Bundled in assets/fonts/ (SIL OFL — free for commercial use):

  • Vazirmatn — the default for everything: body, UI, documents. 9 weights.
  • Lalezar — display font for headlines, covers, posters. One weight; never body.

Other families (Shabnam, Sahel, Samim, Parastoo, Tanha, Gandom) and pairing advice: references/fonts.md. In offline sandboxes only bundled fonts exist — scripts/download_fonts.py works only where GitHub is reachable.

Before any docx→PDF or pptx→PDF conversion, install the fonts:

bash
bash scripts/install_fonts.sh   # copies assets/fonts → ~/.fonts, runs fc-cache
fc-list | grep -i vazir         # verify — else LibreOffice silently falls back

Skipping this is the #1 cause of broken Persian PDFs: conversion "succeeds" but every glyph is DejaVu tofu or disconnected letters.

Documents: the rules that always apply

Format-specific recipes live in the reference files. The universal ones:

  1. RTL means START, not RIGHT. In OOXML, with bidirectional: true, AlignmentType.RIGHT renders at the visual left. Always align START.
  2. Every Persian run: rightToLeft: true + font with hint: "cs" (Complex Script).
  3. Every section: bidi: true. Tables that must flow right-to-left: visuallyRightToLeft: true.
  4. Persian digits in numbered content — a Latin "1." flips the paragraph LTR. Corollary for docx: never use the built-in List Number/List Bullet styles. Their markers live in numbering.xml, which has no bidi and renders Latin 1. on the wrong side plus an OpenSymbol bullet that breaks font embedding. Write markers as ordinary runs («۱. », «• ») — docx-pdf.md §6.1.
  5. Fix the document defaults before adding content. python-docx starts from a Latin template: styles.xml has no Persian font and Word's Heading styles carry a blue color you never asked for. Run persianize_styles(doc) from docx-pdf.md §6.1 immediately after Document(). Headers/footers are separate XML parts and need the RTL treatment applied to them directly.
  6. Pagination: headings get keepNext + keepLines (no orphan headings); cards/boxes go inside a single-cell table with cantSplit: true (never split across pages); no separator after the last list item; no stray PageBreak before a section break (blank pages).
  7. Symbols: Persian fonts miss many glyphs. In bundled Vazirmatn/Lalezar only • and · are verified; ▪ ■ ✓ ✕ ● ◆ ⊙ fall back to DejaVu. For any other symbol/font, check the cmap first (fonts.md shows how).

Verify before delivering

Two gates. Check the .docx BEFORE converting (catches what code review can't), then check the PDF:

bash
# 1. DOCX: package integrity FIRST (a file Word won't open can't be RTL-checked),
#    then section bidi, jc=right traps, cs fonts, list numbering, heading colors.
#    --fix repairs missing <w:bidi/>; --sanitize strips the Word-for-Mac template
#    artifacts python-docx inherits (the classic "file is corrupt" fingerprint).
python3 scripts/verify_docx.py output.docx --expect-font Vazirmatn --fix --sanitize

# 2. PDF: fallback fonts, blank pages, template leaks, Arabic chars
python3 scripts/verify_pdf.py output.pdf --expect-font Vazirmatn

verify_docx.py exists because setting an RTL flag and that flag reaching the XML are different things: libraries drop bidi from section properties, and OOXML requires <w:bidi/> to be the FIRST child of <w:sectPr> — appended anywhere else, renderers ignore it. Verify the artifact, never the source code.

It also guards the package itself: [Content_Types].xml must be the first ZIP entry, every relationship Target must resolve, every XML part must be well-formed and free of control characters. Repairs are written to a temp file and validated before replacing anything, so a failed repair can't destroy the document. Details and the already-corrupt-file procedure: docx-pdf.md §4.5.

It checks: near-empty pages, "undefined"/template leaks, non-embedded or fallback fonts, Arabic ي/ك in extracted text, and page count. Fix every warning, regenerate, re-verify. For prose, re-read your final text against the native test — one pass of persian_cleanup.py --edit + fa_lint.py --check plus one honest read-aloud catches most disasters.

Files

persian-writing/
├── SKILL.md                    ← you are here
├── references/
│   ├── writing-style.md        ← registers, Persian AI tells, colloquial patterns
│   ├── academic.md             ← نگارش علمی: papers, theses, citations, formulas
│   ├── orthography.md          ← ZWNJ, characters, digits, punctuation, ezafe
│   ├── cleanup/                ← paknevis rules, davat API, usage patterns
│   ├── fonts.md                ← catalog, personalities, pairings, embedding
│   ├── content-structures.md   ← POV, education vs promotion, body patterns
│   ├── social-channels.md      ← Telegram/Instagram craft, repurposing
│   ├── seo-copywriting.md      ← Persian SEO writing + کپیرایتینگ
│   ├── docx-pdf.md             ← RTL docx recipes, pagination, PDF post-processing
│   ├── pptx.md                 ← RTL PowerPoint via python-pptx / html2pptx
│   ├── html-css.md             ← RTL web, mixed-direction text, print CSS
│   └── format-skills-fa.md     ← PIL posters, reportlab PDFs, Excel RTL
├── scripts/
│   ├── persian_cleanup.py      ← FIX: edit/clean/normalize/spell-check (paknevis+davat)
│   ├── fa_lint.py              ← LINT: report issues needing contextual judgment
│   ├── verify_docx.py          ← DOCX RTL checks (+ --fix for section bidi)
│   ├── verify_pdf.py           ← post-generation PDF checks
│   ├── check_version.py        ← verify the version matches in all four places
│   ├── install_fonts.sh        ← bundled fonts → ~/.fonts (run before PDF export)
│   └── download_fonts.py       ← fetch extra families (needs GitHub access)
└── assets/
    ├── fonts/                  ← Vazirmatn + Lalezar TTFs
    └── persian_words.txt       ← 453K-word frequency dictionary (spell-check)

© ali2000hos, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. 27 hidden characters (zero-width or bidirectional) removed. Raw file

Files

SKILL.md and 56 other files (scripts, references, assets) in the repository root of ali2000hos/persian-writing.

  • SKILL.md
  • .claude-plugin/marketplace.json
  • .claude-plugin/plugin.json
  • .gitignore
  • AGENTS.md
  • CONTRIBUTING.md
  • LICENSE
  • README.md
  • assets/fonts/Lalezar-Regular.ttf
  • assets/fonts/README.md
  • assets/fonts/Vazirmatn-Black.ttf
  • assets/fonts/Vazirmatn-Bold.ttf
  • assets/fonts/Vazirmatn-Medium.ttf
  • assets/fonts/Vazirmatn-Regular.ttf
  • assets/fonts/Vazirmatn-SemiBold.ttf
  • assets/persian_words.txt
  • assets/persian_words.txt.README
  • docs
  • … and 39 more

Open the folder on GitHubat commit 75fcdf3

Compare with similar skills

Persian Writing next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Persian Writing compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Persian Writing this skillali2000hos/persian-writing368—~4.4kAutomated safety check: WarnMIT
MarkitdownImCa0/just-laws78114 repos~3.2kAutomated safety check: NotesMIT
GenOffice Document CLIgenspark-ai/genoffice9.2k—~19kAutomated safety check: PassApache-2.0
Docx4jplutext/docx4j2.4k—~2.5kAutomated safety check: PassNone
Cyber Pptcrazyykhllc-bit/CyberPPT1.8k—~10kAutomated safety check: PassMIT
Markitshift-labs-ai/markit1.3k—~299Automated safety check: PassMIT

Similar skills

  • Markitdown

    ImCa0/just-laws

    Convert files and office documents to Markdown. An agent skill from ImCa0/just-laws.

    781 GitHub starsUsed in 14 repos~3.2k tokens
    Documents & OfficeAuto-check: notes

  • GenOffice Document CLI

    genspark-ai/genoffice

    Creates, converts, reads and edits real pptx, xlsx, docx and PDF files locally through the genoffice command line.

    9.2k GitHub stars~19k tokensUpdated yesterday
    Documents & OfficeAuto-check passed

  • Docx4j

    plutext/docx4j

    A skill your agent uses when writing Java code that creates, reads or edits Word (.docx), PowerPoint (.pptx) or Excel (.xlsx) files with docx4j — including generating documents, editing existing…

    2.4k GitHub stars~2.5k tokensUpdated 2 days ago
    Documents & OfficeAuto-check passed

  • Cyber Ppt

    crazyykhllc-bit/CyberPPT

    当用户需要把 DOCX、PDF、TXT、XLSX、研究报告、业务材料或原始数据转成高密度、可编辑、咨询风格 PPTX 时使用；也适用于需要 SCR 论证、视觉风格探索、详细图表和渲染质检的 PPT。

    1.8k GitHub stars~10k tokensUpdated 2 mo ago
    Documents & OfficeAuto-check passed

  • Markit

    shift-labs-ai/markit

    Convert files and URLs to Markdown. An agent skill from shift-labs-ai/markit.

    1.3k GitHub stars~299 tokensUpdated 1 mo ago
    Documents & OfficeAuto-check passed

  • Markitdown

    jimmc414/Kosmos

    Convert various file formats (PDF, Office documents, images, audio, web content, structured data) to Markdown optimized for LLM processing.

    595 GitHub starsUsed in 2 repos~1.7k tokens
    Documents & OfficeAuto-check passed

Works with

Categories

Questions about Persian Writing

What does Persian Writing do?

Write natural, human-sounding Persian (Farsi) and build correct right-to-left Persian deliverables. Persian Writing is an agent skill from ali2000hos/persian-writing. Write natural, human-sounding Persian (Farsi) and build correct right-to-left Persian deliverables.

When should I use Persian Writing?

Persian Writing fits situations like: ANY task involving Persian text; an Iranian audience: writing; SEO content and copywriting (سئو، کپیرایتینگ، لندینگ); academic papers/theses.

How do I install Persian Writing in Claude Code?

Run `npx skills add ali2000hos/persian-writing --skill persian-writing -a claude-code`. Or copy the skill folder (the ali2000hos/persian-writing repository) into .claude/skills/persian-writing in your project. Claude Code loads it when a task matches its description.

How do I install Persian Writing in Codex?

Run `npx skills add ali2000hos/persian-writing --skill persian-writing -a codex`. Or copy the skill folder (the ali2000hos/persian-writing repository) into .agents/skills/persian-writing in your project. Codex loads it when a task matches its description.

Can I use Persian Writing in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add ali2000hos/persian-writing --skill persian-writing -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/persian-writing, .gemini/skills/persian-writing, .github/skills/persian-writing and .opencode/skills/persian-writing in your project.

What does Persian Writing need to run?

Going by SKILL.md and its folder, Persian Writing needs the command-line tools its instructions call (python3, bash and npm). Our summary lists: Python 3; Node.js; Docker. Compatibility (from SKILL.md): Any agent that reads Markdown skills (Claude, Claude Code, Cursor, Codex, custom harnesses; chat-only AIs via universal/persian-writing-universal.md). Scripts: Python 3.8+ stdlib only. docx/pptx→PDF conversion: LibreOffice + installed fonts. PDF checks: poppler-utils, pypdf, or PyMuPDF (any one). .

Does Persian Writing access the network?

SKILL.md contains no URLs. Its commands use npm, which can reach the network depending on how they are called. This is read from the text; nothing was executed.

Is Persian Writing safe to install?

Our automated static check of SKILL.md flagged 10 warning(s): contains zero-width characters. Read the flagged lines before installing; the check is not a guarantee either way. The check reads SKILL.md only: the scripts in the folder are not scanned, so read them before running anything.

What licence does Persian Writing use?

Persian Writing is published under the MIT licence (declared in SKILL.md). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Persian Writing use?

About 4.4k tokens (SKILL.md is roughly 17k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 42k tokens, read only when the agent opens those files.

What are the alternatives to Persian Writing?

Skills that share tags, products or a category with Persian Writing: Markitdown (ImCa0/just-laws, 781 stars), GenOffice Document CLI (genspark-ai/genoffice, 9.2k stars), Docx4j (plutext/docx4j, 2.4k stars) and Cyber Ppt (crazyykhllc-bit/CyberPPT, 1.8k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Persian Writing?

ali2000hos (a GitHub user) maintains it in ali2000hos/persian-writing, which has 368 GitHub stars. The repository was last updated on October 4, 2026.

Source: ali2000hos/persian-writing on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.