Quality Evaluation Loop

I score any deliverable before it is returned: accuracy and sources, Arabic quality and dialect, RTL formatting, brand compliance, completeness, and unfinished placeholders. I revise for up to three rounds and log the scores.

What it produces

• Scores on six axes
• Text after revision
• A log of the rounds

Questions it asks when woken

• Paste the deliverable to score
• What was the original objective?

Agents in the same department