Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add ahmeddabak/claude-arabic-writing --skill arabic-writinggit clone --depth 1 https://github.com/ahmeddabak/claude-arabic-writingWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/ahmeddabak/claude-arabic-writing/arabic-writing)<a href="https://agentmods.dev/skills/ahmeddabak/claude-arabic-writing/arabic-writing"><img src="https://agentmods.dev/badge/skills/ahmeddabak/claude-arabic-writing/arabic-writing/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/ahmeddabak/claude-arabic-writing/arabic-writing"><img src="https://agentmods.dev/badge/skills/ahmeddabak/claude-arabic-writing/arabic-writing.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00117 | $0.01985 |
| Opus 5.5 | $0.00047 | $0.00794 |
| Sonnet 5.5 | $0.00023 | $0.00397 |
| Haiku 4.5 | $0.00012 | $0.00198 |
Grade A, and why
arabic-writing scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 107 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Writing natural, correct Arabic
This skill helps you produce Arabic that a native editor would accept: grammatically correct, idiomatic, free of typos, and free of the tell-tale signs of machine translation. The single biggest failure mode is translationese — transferring English structure word-for-word instead of re-composing the meaning in Arabic. Translate ideas, not words.
Workflow
- Pick the register and dialect. Default to Modern Standard Arabic (MSA / الفصحى) for anything formal: news, documents, legal/medical text, UI body copy, subtitles. Use a dialect (Egyptian, Levantine, Gulf, Maghrebi) or light "white Arabic" only when the context is conversational, marketing, social, or the user asks for it. When unsure, ask which dialect/market — never silently mix dialect into formal MSA. See register-and-style.md.
- Compose, don't transfer. Read the whole source sentence, understand the meaning, then write it the way an Arabic speaker would — usually a verb-first sentence with the right collocation — rather than mapping English word order and prepositions.
- Self-check every output against the two checklists below before returning it.
- Open the relevant reference file whenever you are unsure about a specific rule. The reference files hold the detail; this page is just the map.
Top correctness checklist (apply to every Arabic output)
- Non-human plurals take feminine-singular agreement. الكتبُ جديدةٌ ("the books are new"), السياراتُ سريعةٌ — not جُدُد/سريعات. This is the #1 error. (→ grammar.md)
- Verb–subject order changes agreement. Verb before subject (VSO, the neutral order) stays singular and agrees only in gender: ذهب الطلابُ / ذهبت الطالباتُ. Subject before verb (SVO) takes full agreement: الطلابُ ذهبوا. (→ grammar.md)
- Iḍāfa (الإضافة): the first noun (مضاف) takes no ال and no tanwin; the second is genitive. كتابُ الطالبِ, not الكتابُ الطالبِ. No adjective between the two parts. (→ grammar.md)
- Adjectives follow the noun and agree on all four axes: gender, number, case, definiteness. البنتُ الكبيرةُ (both definite) vs البنتُ كبيرةٌ ("the girl is big"). (→ grammar.md)
- Numbers 3–10 reverse gender (polarity) and take a plural genitive noun: ثلاثةُ رجالٍ, ثلاثُ فتياتٍ. 11–99 take a singular accusative (tamyīz): عشرون كتابًا. (→ grammar.md)
- Hamza sits on the seat of the strongest vowel (kasra ئ > damma ؤ > fatha أ > sukūn ء): رئيس، مسؤول، سأل، شيء. Distinguish wasl (bare ا) from qatʿ (أ/إ) with the و-test. (→ orthography.md)
- ة vs ه (add a suffix: /t/ → ة, /h/ → ه) and ى vs ي (trace the root; على the preposition vs عليّ the name). (→ orthography.md)
- Use Arabic punctuation glyphs ، ؛ ؟ (not , ; ?). No space before a mark, one after. In lists, repeat و instead of using commas. (→ orthography.md)
- Default to undiacritized text; add ḥarakāt only to remove genuine ambiguity, but always keep a meaning-distinctive shadda (درس vs درّس). Never use kashida (ـــ) to pad. (→ orthography.md)
- Right collocations and prepositions. اتخذ قرارًا (not عمل قرارًا), أثّر في (not على in formal MSA), حصل على. (→ collocations.md, translationese.md)
What ships with it
11 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- evals/evals.json 4.8 KB
- references/collocations.md 6.0 KB
- references/conventions.md 4.6 KB
- references/dialects/egyptian.md 2.7 KB
- references/dialects/gulf.md 2.7 KB
- references/dialects/levantine.md 2.5 KB
- references/dialects/maghrebi.md 3.1 KB
- references/grammar.md 7.6 KB
- references/orthography.md 6.5 KB
- references/register-and-style.md 5.3 KB
- references/translationese.md 8.0 KB
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2mo ago First seen · 107 lines · 117 tokens per session scan A 5f25d2f5307c
arabic-writing is a skill published in the GitHub repository ahmeddabak/claude-arabic-writing (20 stars, last pushed 4mo ago), licensed MIT. It adds 117 tokens to every session and 1,985 once invoked, about $0.0005 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
arabic-writing compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| arabic-writingahmeddabak/claude-arabic-writing | Skill | 20 | 4mo ago | MIT | A | 117 | Claude Code, Codex |
| arabic-copy-localizergrowthack88/growth-marketing-os | Skill | 127 | 27d ago | MIT | A | 94 | Claude Code |
| arabic-word-productionBannovich/arabic-word-production | Skill | 10 | 1mo ago | Apache-2.0 | A | 38 | Codex |
| arabicskilltheonlym7md/arabic-ai-skills | Skill | 2 | 4d ago | MIT | A | 66 | Codex |
| arabic-uiux-mastertheonlym7md/arabic-ai-skills | Skill | 2 | 4d ago | MIT | A | 60 | Codex |
| auto-reader-ocrSamqra/auto-reader-ocr | Skill | 0 | 3mo ago | MIT | B | 101 | Claude Code |
| arabic-brand-voicesmeseik-ai/arabic-content-studio | Skill | 5 | 7mo ago | none | A | 115 | Claude Code, Codex |