A
Ariiima/persian-writer
Claude skill that writes natural, native-sounding Persian (Farsi) text
View on GitHub ↗Link checked 2026-11-12 · HTTP 200 · opens the vendor's own site
- Star count
- 1 ★ · Ariiima/persian-writer · read from the GitHub search API · (checked 2026-11-13 20:00:50)
- Source
- read directly from its GitHub page · type: open-source project
- Harvest history
- first seen 2026-10-24 · last seen 2026-11-14 · seen 25× · listed by 1 sources
- Score
- score 103 · 1 ★ · 1 sources · repo
FAQ (answers come from our records)
Who is it for?
We filed it under these tags: 开源项目. The tags come from the entry itself, not a taxonomy we imposed.
Who maintains it?
repository
Ariiima/persian-writer；first catalogued 2026-10-24；last seen 2026-11-14。There is no official certification or quality rating here — we record sources and dates only.# persian-writer
以下摘录自该仓库的 README（片段） · 在 GitHub 读完整 README →
نگارش فارسی طبیعی و استاندارد برای دستیارهای هوش مصنوعی — Natural, native-sounding Persian (Farsi) writing skill for Claude, Claude Code, Cursor & Codex
Large language models know Persian grammar rules on paper, yet their Persian output routinely reads as machine-translated (ترجمه‌زده) or stiffly bureaucratic (اداری‌زده). This skill fixes the four root causes without bloating context with .docx / .pptx / LibreOffice layout instructions:
1. Syntactic Anti-Calque (ضد گرته‌برداری نحوی): Shifts English noun-heavy, passive (توسط), SVO-ordered sentences into verb-centric Persian.
2. Bureaucratic & AI Fossil Removal (حذف کلیشه‌های اداری و ماشینی): Strips می‌باشد, گردید, نمود, در راستای, لازم به ذکر است, مورد ... قرار گرفتن, نه تنها... بلکه همچنین, and em-dashes (—).
3. 5-Register Routing & Lock (تشخیص و قفل رجیستر): Separates the user's prompt tone from the deliverable register and prevents mid-paragraph mixing between written (کتابی) and spoken (شکسته) forms.
4. Farhangestan 1401 Orthography (دستور خطّ فرهنگستان ۱۴۰۱): Enforces ZWNJ (نیم‌فاصله) in both formal and colloquial verbs (می‌شود / می‌شه), prevents هکسره errors (کتابه من ❌ → کتابِ من ✅), preserves silent ه in colloquial clitics (آماده‌ست, نتیجه‌ش), bans fake tanvin (گاهاً → گاهی, دوماً → دوم اینکه), and normalizes digits/punctuation.
## Supported Registers
| Register | Surface / Audience | Verbs | Object & Pronouns |
| formal-human (رسمی روان — Default) | B2B emails, proposals, reports, docs, standard UI, company websites | است, شد, می‌شود, می‌توانید | را, این / آن, آن‌ها, شما |
| formal-admin (رسمی اداری) | Official letters (نامه اداری) to ministries, tax office, banks, universities | است, می‌شود + دستور فرمایید | را, این / آن, جناب‌عالی |
| semi-formal (نیمه‌رسمی / آموزشی) | Blog posts, newsletters, tutorials, technical guides, LinkedIn | است, می‌شود, می‌کنیم | را, این / آن, direct شما |
| light-colloquial (محاورهٔ نرم) | Friendly consumer app UI, onboarding, casual landing pages, support chat | می‌شه, می‌تونید, داره, ـه | Keeps را and این / آن |
| full-colloquial (محاورهٔ کامل / نریشن) | Instagram/Telegram captions, friendly DMs, podcast & video voiceover (نریشن) | می‌شه, می‌خوام, می‌رم, می‌گه, بذار | رو, اون / اونا / اینا, واسه, دیگه |
## Repository Structure
text
persian-writer/
├── SKILL.md # Core router, register lock table, anti-calque rules & verification workflow
├── references/
│ ├── registers.md # 6-step register detection, lock table & rules per register
│ ├── anti-calque-and-style.md # English→Persian syntactic transforms, calque lexicon & AI-tell fixes
│ ├── orthography-1401.md # Farhangestan 1401 ZWNJ rules, هکسره differential test, tanvin, BiDi & numbers
│ └── ui-and-microcopy.md # Product UX copy, buttons, error messages, SMS & landing pages
├── scripts/
│ ├── facheck.py # Zero-dependency deterministic linter, register-mix detector & safe auto-fixer
│ └── runbenchmark.py # Automated evaluation harness (facheck + blind pairwise LLM judge)
├── evals/
│ ├── evals.json # 8 golden evaluation tasks across all 5 registers
│ ├── triggers.json # 20 trigger / non-trigger test queries
│ └── files/ # Sample AI-generated & orthographically broken test inputs
Large language models know Persian grammar rules on paper, yet their Persian output routinely reads as machine-translated (ترجمه‌زده) or stiffly bureaucratic (اداری‌زده). This skill fixes the four root causes without bloating context with .docx / .pptx / LibreOffice layout instructions:
1. Syntactic Anti-Calque (ضد گرته‌برداری نحوی): Shifts English noun-heavy, passive (توسط), SVO-ordered sentences into verb-centric Persian.
2. Bureaucratic & AI Fossil Removal (حذف کلیشه‌های اداری و ماشینی): Strips می‌باشد, گردید, نمود, در راستای, لازم به ذکر است, مورد ... قرار گرفتن, نه تنها... بلکه همچنین, and em-dashes (—).
3. 5-Register Routing & Lock (تشخیص و قفل رجیستر): Separates the user's prompt tone from the deliverable register and prevents mid-paragraph mixing between written (کتابی) and spoken (شکسته) forms.
4. Farhangestan 1401 Orthography (دستور خطّ فرهنگستان ۱۴۰۱): Enforces ZWNJ (نیم‌فاصله) in both formal and colloquial verbs (می‌شود / می‌شه), prevents هکسره errors (کتابه من ❌ → کتابِ من ✅), preserves silent ه in colloquial clitics (آماده‌ست, نتیجه‌ش), bans fake tanvin (گاهاً → گاهی, دوماً → دوم اینکه), and normalizes digits/punctuation.
## Supported Registers
| Register | Surface / Audience | Verbs | Object & Pronouns |
| formal-human (رسمی روان — Default) | B2B emails, proposals, reports, docs, standard UI, company websites | است, شد, می‌شود, می‌توانید | را, این / آن, آن‌ها, شما |
| formal-admin (رسمی اداری) | Official letters (نامه اداری) to ministries, tax office, banks, universities | است, می‌شود + دستور فرمایید | را, این / آن, جناب‌عالی |
| semi-formal (نیمه‌رسمی / آموزشی) | Blog posts, newsletters, tutorials, technical guides, LinkedIn | است, می‌شود, می‌کنیم | را, این / آن, direct شما |
| light-colloquial (محاورهٔ نرم) | Friendly consumer app UI, onboarding, casual landing pages, support chat | می‌شه, می‌تونید, داره, ـه | Keeps را and این / آن |
| full-colloquial (محاورهٔ کامل / نریشن) | Instagram/Telegram captions, friendly DMs, podcast & video voiceover (نریشن) | می‌شه, می‌خوام, می‌رم, می‌گه, بذار | رو, اون / اونا / اینا, واسه, دیگه |
## Repository Structure
text
persian-writer/
├── SKILL.md # Core router, register lock table, anti-calque rules & verification workflow
├── references/
│ ├── registers.md # 6-step register detection, lock table & rules per register
│ ├── anti-calque-and-style.md # English→Persian syntactic transforms, calque lexicon & AI-tell fixes
│ ├── orthography-1401.md # Farhangestan 1401 ZWNJ rules, هکسره differential test, tanvin, BiDi & numbers
│ └── ui-and-microcopy.md # Product UX copy, buttons, error messages, SMS & landing pages
├── scripts/
│ ├── facheck.py # Zero-dependency deterministic linter, register-mix detector & safe auto-fixer
│ └── runbenchmark.py # Automated evaluation harness (facheck + blind pairwise LLM judge)
├── evals/
│ ├── evals.json # 8 golden evaluation tasks across all 5 registers
│ ├── triggers.json # 20 trigger / non-trigger test queries
│ └── files/ # Sample AI-generated & orthographically broken test inputs
Listed badge (optional)
Copy this line into the README if you want the listing mark. Optional, and it does not affect being listed.
[](https://anclum.com/en/s/ariiima-persian-writer/)
- Source
- github · https://github.com/Ariiima/persian-writer
- Browse by field
- Open source
- Traction
- 1 ★
- added
- today
More like this
Filed under
Every tag opens the full list for that filter. Counts are the live catalogue figures.