🏷️ Tag Library

Use tags as performance cues, not decoration. The safest batch tags are lower-case, sparse, and placed only when delivery changes.

🎭 Core Set

• [announcer]: show-host posture.
• [suspenseful]: quiz stakes.
• [clearly]: intelligible answer list.
• [curious]: inviting option scan.
• [pause]: short decision beat.
• [inviting]: login/action prompt without pressure.
• [confident]: correct-answer reveal.
• [warm]: human, non-robotic tone.
• [pleased]: positive continuation.
• [encouraging]: gentle next-step energy.

🎚️ Combos

• prompt_open: [announcer, suspenseful].
• options: [clearly, curious].
• decision_gate: [pause] then [inviting].
• reveal: [confident, warm].
• continuation: [pleased, encouraging].

🚫 Avoid

• heavy SFX in canonical quiz questions.
• accents unless a voice test proves stable.
• sarcasm on factual answers.
• more than two tags in one bracket.
• tags between A/B/C/D options, because they can accidentally favor one answer.

📈 Ladder

• level_1: plain text plus punctuation; most reliable for facts.
• level_2: one tag at a section boundary.
• level_3: two compatible tags in one bracket.
• level_4: persona/reaction tags; useful for variants.
• level_5: heavy SFX, accents, or dense stacks; reserve for experiments.

✅ Batch Rule

The canonical WWTBAM batch uses level 3 at open/reveal boundaries and level 2 for the decision gate. It avoids SFX so every item remains fair, repeatable, and easy to validate.