---
name: content-quality-rubric
description: "Foundation skill — always load alongside brand-voice-smorchestra. The 8-dimension scoring rubric exposed to the writer at generation time. Self-correct before the scorer ever sees the output. Minimum 9/10 to ship. Nothing leaves without a score table."
---

# Content Quality Rubric

## The Rule
Score every post before delivering. Show the table. Show path to 10/10. Nothing ships below 9/10.

---

## 8 Dimensions

| # | Dimension | Weight | 10/10 means |
|---|---|---|---|
| 1 | Hook Strength | 20% | Specific number OR bold contrarian claim. Reader cannot NOT read on. |
| 2 | Narrative Coherence | 15% | Every line earns its place. Hook → tension → resolution → CTA. No filler. |
| 3 | Proof Density | 15% | Every claim has a number, name, or case. "$180K," "Riyadh," "90 days." |
| 4 | Brand Voice Match | 15% | Indistinguishable from Mamoun's real posts in voice-match-9plus.md |
| 5 | CTA Effectiveness | 10% | DM keyword CTA + P.S. question. Both present. |
| 6 | Audience Relevance | 10% | MENA-specific. Arab founder context. Not generic B2B. |
| 7 | Language Quality | 10% | Zero forbidden phrases. No AI tells. Reads naturally. |
| 8 | Differentiation | 5% | Could not be posted by a competitor without major edits. |

---

## Scoring Formula
```
Final = (Hook×0.20) + (Narrative×0.15) + (Proof×0.15) + (Voice×0.15) +
        (CTA×0.10) + (Audience×0.10) + (Language×0.10) + (Differentiation×0.05)
```

---

## Score Table Format
| Dimension | Weight | Score | Issue |
|---|---|---|---|
| Hook strength | 20% | 9 | |
| Narrative coherence | 15% | 8 | Section 2 padded |
| Proof density | 15% | 9 | |
| Brand voice match | 15% | 7 | 3 AI phrases |
| CTA effectiveness | 10% | 9 | P.S. missing |
| Audience relevance | 10% | 9 | |
| Language quality | 10% | 8 | "leverage" appears |
| Differentiation | 5% | 9 | |
| **Weighted average** | | **8.55** | **Fix before shipping** |

---

## Thresholds
| Score | Action |
|---|---|
| 9.0+ | Ship ✅ |
| 8.5-8.9 | Fix 1-2 lowest dimensions, rescore |
| 8.0-8.4 | Rewrite 3+ dimensions |
| Below 8.0 | Regenerate from scratch |

---

## Self-Correction Protocol
1. Read only lines 1-2. Would you stop scrolling? If not — rewrite hook first.
2. Run forbidden-phrases-en.md or forbidden-phrases-ar.md. Delete all matches.
3. Highlight every claim. Add a number or name to any claim without one.
4. Read aloud. Compare to voice-match-9plus.md. Same person?
5. Is there a DM CTA with keyword? P.S. question? If not — add them.
6. Score. If below 9.0 — fix and rescore before delivering.

See examples/ for annotated score breakdowns at 9.2, 8.5, and 7.5.