Persian Language

Enhances AI ability to read, write, translate, and format Persian (Farsi) with native-level accuracy across any task. Handles Unicode, half-spaces, RTL, registers, ta'arof, Solar Hijri dates, and cultural nuance.

Persian Language Skill

Identity

A capability layer that enhances the agent's ability to read, write, translate, and format Persian (Farsi) across any task. Not a tutor. Not a chatbot persona. A quality multiplier for any workflow involving Persian.

---

Triggers

Activate this skill when any of the following are true:

• The user writes in Persian (Farsi script)
• The user requests Persian content generation (posts, emails, docs, reports, stories, comments)
• The user asks to translate to or from Persian
• The user asks to review, improve, or edit Persian text
• Persian text appears in an attached file, image, or code
• The user mentions "فارسی", "Persian", or "Farsi" in a task
• Mixed Persian/English content is present and needs handling

---

Core Instructions

1. Always Use Correct Persian Unicode

• Use ک (U+06A9 Persian kaf), never ك (U+0643 Arabic kaf)
• Use ی (U+06CC Persian ya), never ي (U+064A Arabic ya)
• Use half-space (U+200C) in compound words: می‌خواهم، نمی‌توانم، خانه‌ام
• See references/writing-standards.md for full rules

2. Format Persian Output Correctly

• Use Persian punctuation: «» for quotes, ، for comma, ؛ for semicolon, ؟ for question mark
• Default to Persian digits (۰۱۲۳۴۵۶۷۸۹, U+06F0–U+06F9) in prose — not Arabic-Indic digits (٠١٢٣٤٥٦٧٨٩, U+0660–U+0669), which look similar but are wrong in Persian
• Use Persian numeric separators: ٬ (U+066C) for thousands, ٫ (U+066B) for decimal, ٪ (U+066A) for percent — never , . or % inside a Persian number
• Dates in Persian content use the Solar Hijri calendar (شمسی), e.g. ۲۴ فروردین ۱۴۰۵
• Keep Western digits in code, technical IDs, versions, ports, URLs, and anything inside a code block
• Respect right-to-left (RTL) text direction — do not let punctuation or Latin fragments break flow, and never manually reverse digits to "fix" their display
• In mixed content, isolate LTR segments properly
• See references/numerals.md for the full numerics guide (digit families, separators, dates, time, currency, percentages, ordinals, phone numbers, math, mixed content)

3. Match the Right Register

• Formal (رسمی): reports, business emails, academic writing, official announcements — use شما, formal verb endings, no contractions
• Colloquial (محاوره‌ای): social media, casual messages, dialogue — Tehran-standard spoken forms are acceptable
• Mixed/code-switching: when Persian text includes English technical terms, integrate them naturally without forced translation of well-known terms (e.g., API, framework, deploy)
• See references/tone-register.md for register details

4. Translate with Cultural Nuance

• Persian → English: preserve the tone — formal stays formal, sarcastic stays sarcastic, ta'arof is explained or adapted, not dropped
• English → Persian: choose the natural Persian expression, not a word-for-word calque
• Idioms: translate the meaning, not the words — provide the original if helpful
• See references/common-mistakes.md for translation pitfalls

5. Handle Cultural Context

• Ta'arof: recognize that excessive politeness in Persian is often formulaic, not literal. "قابلی نداره" does not mean the item has no value.
• Dates: Iran uses the Solar Hijri calendar (شمسی/هجری خورشیدی). When dates matter, provide شمسی alongside Gregorian. Current year: ۱۴۰۵ هجری خورشیدی.
• Names: Persian names may include titles (آقای، خانم، دکتر، مهندس) — preserve them when appropriate.

6. Maintain Quality Across Tasks

This skill is not limited to one use case. Apply Persian capabilities to:

• Content writing (blog posts, captions, ad copy)
• Document drafting (formal letters, reports, proposals)
• Code comments and documentation in Persian
• Data extraction from Persian text
• Summarization of Persian sources
• Localization and adaptation of English content for Iranian audiences

---

Reference Files

File	Purpose
references/writing-standards.md	Unicode, punctuation, numerals, RTL formatting
references/numerals.md	Digit families, separators, dates, time, currency, percentages, ordinals, phone, math, mixed content
references/tone-register.md	Formal/informal, ta'arof, politeness, greetings
references/common-mistakes.md	AI error patterns in Persian + corrections
references/transliteration.md	Standard romanization when Latin script is needed
references/content-templates.md	Ready-made templates: email, social, report, announcement

---

Quick Examples

Unicode: Right vs Wrong

❌ Bad (Arabic Unicode):

text

كتاب - ي - نمي خواهم

✅ Good (Persian Unicode):

text

کتاب - ی - نمی‌خواهم

Register Matching

❌ Formal email with informal ending:

text

با احترام،
موضوع جلسه رو بررسی کردیم.
مرسی!

✅ Consistent formal register:

text

با احترام،
موضوع جلسه را بررسی کردیم.
با تشکر و احترام

Half-Space Usage

❌ Missing half-spaces:

text

نمیتوانم کتابها را بخوانم

✅ Correct half-spaces:

text

نمی‌توانم کتاب‌ها را بخوانم

---

Testing This Skill

Test with these prompts:

• "Write a formal Persian email about a meeting"
• "Translate this to Persian: The project deadline is next Monday"
• "Fix this Persian text: كتاب را نمي خواهم"
• "Create a Persian Instagram caption for a sunset photo"
• "Summarize this article in Persian: [paste English text]"

---

Quality Checklist (Apply Before Returning Persian Output)

• No Arabic Unicode characters (ك → ک, ي → ی)
• Half-spaces used in compound words (می‌، نمی‌، ها، ترین)
• Persian punctuation used (« » ، ؛ ؟)
• Register is consistent (not mixing formal and colloquial)
• Numbers match context: Persian digits (U+06F0–U+06F9) in prose, Western in code/technical
• No Arabic-Indic digits (U+0660–U+0669) — check ۴/٤, ۵/٥, ۶/٦ especially
• Separators inside numbers are ٬ (thousands), ٫ (decimal), ٪ (percent) — not , . %
• Dates use Solar Hijri (شمسی); calendar is signaled when mixing with Gregorian
• RTL formatting is intact — no broken punctuation or misplaced Latin fragments
• Translation reads naturally, not like a calque
• Cultural references are accurate and current