Arabic Content Quality

Use this skill as the Arabic language, Unicode, RTL, and compatibility layer on top of the skill for the requested output format. Preserve the user's content, design, and target format.

Core contract

• Store Arabic in normal logical Unicode order. Never reverse characters, words, or lines and never pre-shape letters into Arabic Presentation Forms.
• Do not add invisible bidi controls (RLE, LRE, RLO, LRO, PDF, RLI, LRI, FSI, PDI, LRM, RLM, or ALM) as a general RTL fix. Use native paragraph, container, or document direction.
• Do not delete ZWJ or ZWNJ blindly; they may be linguistically meaningful. Flag unexpected occurrences and remove them only when safe.
• Set Arabic language metadata (ar-SA unless the requested locale differs), native RTL direction, and intentional alignment independently.
• Keep long English passages, URLs, email addresses, file paths, citations, formulas, and code in separate LTR containers when the format permits.
• Choose a font with Arabic glyph coverage. Do not claim a font is embedded or portable unless verified. Supply PDF when fixed appearance matters.
• Use native list and numbering features. Do not fake bullets, indentation, tables, or reading order with spaces or typed glyphs.
• Preserve editable source files whenever editability is requested; PDF is a fixed-output companion, not a replacement.

Read references/unicode-and-writing.md whenever authoring or repairing Arabic text. Then read only the reference for the target format: