---
name: ai-shot-list-generator
description: >-
  Create production-ready shot lists from scripts, scenes, advertisements and
  video ideas. Use when the user asks for a shot list, shooting plan, camera
  breakdown, visual breakdown, video scene plan or script-to-shots conversion.
  Include shot size, angle, movement, action, duration, sound and continuity notes.
---

# AI shot list generator

Turn a script, scene, ad brief, or video idea into a camera-ready shot list a DP, creator, or AI video pipeline can execute.

Load reference files under `references/` when you need shot grammar, continuity checks, or use-case templates. Prefer the CSV templates under `templates/` when the user wants a blank sheet. Run `scripts/validate-shot-duration.py` when durations must sum to a target runtime.

---

## Production method (always)

1. Analyse the concept and target audience.
2. Establish runtime and pacing from the target duration.
3. Break the concept into narrative beats.
4. Convert each beat into shots.
5. Vary shot sizes and camera grammar.
6. Check spatial and visual continuity.
7. Check total duration against the target.
8. Add sound and dialogue / voice-over.
9. Generate model-ready video prompts per shot.
10. Validate that each shot can actually be produced with the chosen method (live-action, AI, hybrid, animation).

---

## Inputs to gather

1. **Script, scene, or concept**
2. **Video type** — YouTube video, YouTube Short, UGC ad, product ad, short film, dialogue scene, music video, AI-generated film, documentary, explainer
3. **Target duration**
4. **Aspect ratio**
5. **Visual style**
6. **Number of shots** (or let pacing decide)
7. **Production method** — live-action / AI-generated / hybrid / animation

If inputs are thin, invent reasonable staging and label assumptions.

---

## Output columns

| Field | Purpose |
|-------|---------|
| Shot number | Production order |
| Duration | Pacing and total runtime |
| Shot size | Wide, medium, close-up, insert, etc. |
| Camera angle | Eye-level, low, high, overhead, etc. |
| Camera movement | Static, pan, dolly, handheld, etc. |
| Subject | Person or object being filmed |
| Action | What happens in the shot |
| Composition | Placement and visual emphasis |
| Dialogue / VO | Spoken content |
| Sound | Ambience, effects, music cue |
| Continuity | Props, position, wardrobe, eyeline |
| Transition | Cut, match cut, dissolve, movement |
| AI video prompt | Model-ready generation prompt |

Deliver as a markdown table by default. Offer CSV / JSON if the user asks.

---

## Use-case overrides

- **YouTube:** hook → talking head → B-roll → demo → pattern interrupt → CTA
- **UGC ads:** hook → problem → product CU → usage → proof → offer → CTA
- **Short film:** geography → eyelines → coverage → inserts → emotional CU → continuity → transitions
- **AI film:** character ref → location state → camera → start/end frame → motion → prompt consistency → clip transitions

Read `references/youtube-shorts.md`, `references/ugc-ads.md`, or `references/ai-video-prompts.md` when relevant.

---

## Rules

- One primary action per shot.
- Concrete camera language only.
- Protect continuity across chained shots.
- Match the medium (live vs AI vs hybrid).
- Do not invent brand claims.
- Sum of durations should land within ~15% of the target runtime.

---

## Checklist

- [ ] Beats cover the full concept
- [ ] Shot sizes and angles vary
- [ ] Continuity notes present between related shots
- [ ] Durations sum near the target
- [ ] Sound / dialogue filled where useful
- [ ] AI prompts are consistent when production is AI or hybrid
- [ ] Assumptions labeled when inputs were thin