audit-short-ffa
Skill1 starsChanged 5 months agoMITView on GitHub
Audit pré-publication YouTube d'un short {{GAME_PROJECT}} format 4-way FFA (1V1V1V1). Score viralité 0-100 + verdict GO/WARN/NO-GO + recommandations basées sur algo YouTube Shorts 2026 + forensic du channel {{YT_CHANNEL}} (5 métriques pondérées : durée 35%, iconic personas 30%, action density 20%, hook variance 15%, bonus multi_fighter visible). Seuils FFA différents du 1v1 : durée cible 45-75s (pas 25-45), action density seuil 18 (pas 14), iconic ≥3/4 pour PASS. Use when about to upload a FFA short, or when user asks "audit ce short FFA" / "score viralité FFA" pour un .mp4 4-way.
In this repository11 other files it gives its agents.
- anti-ai-detect-writingSkill
- audit-short-1v1Skill
- awesome-design-mdSkill
- claude-councilSkill
- deep-researchSkill
- design-mdSkill
- design-audit-paris-nuitSkill
- gemini-deep-researchSkill
- gemini-spriteSkill
- oracle-ampere-a1-setupSkill
- smart-routerSkill
The file
Files
Skill : audit-short-ffa
Audit pré-publication d'un short {{GAME_PROJECT}} format 4-way FFA (1V1V1V1). Score 0-100 + verdict GO/WARN/NO-GO + recommandations actionnables.
Pourquoi des seuils différents du 1v1
- Durée cible 45-75s (pas 25-45s) : 4 fighters = besoin de temps pour que chacun agisse. <30s = massacre unilatéral (bug batch 21 Football GOAT 13s → NO-GO).
- Action density ≥18 (pas ≥14) : 4 fighters devraient produire plus de mouvement qu'un duel.
- Iconic ≥3/4 (pas ≥2/2) : FFA a besoin de plus d'icônes pour maximiser l'audience de plusieurs fandoms.
Prérequis
cd ~/Desktop/battle-engine
source .venv/bin/activate
# Deps : cv2, imagehash, pytesseract (déjà installés)Invoquer
cd ~/Desktop/battle-engine
source .venv/bin/activate
python3 pipeline/audit_cli.py --preset ffa /path/to/video.mp4
# Avec matchup explicite (recommandé pour FFA, souvent 4 fighters) :
python3 pipeline/audit_cli.py --preset ffa /path/to/video.mp4 --matchup goku,mario,sonic,pikachuExit code : 0 si GO ou WARN, 1 si NO-GO.
Métriques et poids
| Métrique | Poids | PASS | WARN | FAIL |
duration_s | 35% | 45-75s | 30-44s ou 76-90s | <30s ou >90s |
iconic_score | 30% | ≥3/4 icônes | 2/4 | ≤1/4 |
action_density | 20% | ≥18 | 12-17 | <12 |
hook_variance_t2 | 15% | ≤900 | 901-1500 | >1500 |
Icônes : Mario, Pikachu, Sonic, Goku, Steve (Minecraft).
Bonus : +5 pts si ≥3 fighters distincts visibles à t=1.5s (connected components).
Hard gates NO-GO : résolution != 1080x1920, black_frame_ratio > 10%, freeze_ratio > 15%, durée > 90s.
Interpréter le score
| Score | Verdict | Action |
| 60-100 | GO — top tier probable | Uploader |
| 40-59 | WARN — mid tier | Envisager re-render ou uploader quand même |
| 0-39 | NO-GO — sous-performance attendue | Re-render avec autre seed ou changer matchup |
Que faire si NO-GO
- duration_s FAIL (<30s) : Massacre unilatéral probable (1 fighter trop fort). Re-render impératif. Équilibrer les stats ou choisir des fighters plus proches en puissance.
- duration_s FAIL (>90s) : Exclu de l'algorithme Shorts. Réduire HP ou utiliser comme long format uniquement.
- iconic_score FAIL (<2) : Remplacer 2+ persos par Mario/Pikachu/Sonic/Goku/Steve. Pour FFA l'audience attend des icônes gaming reconnues.
- action_density FAIL (<12) : FFA statique. 4 fighters bloqués ou pas assez agressifs. Re-render avec autre seed.
- hook_variance_t2 FAIL (>1500) : Bug d'intro FFA. Spawn incorrect des 4 fighters. Re-render.
Exemple output JSON (Football GOAT 13s — NO-GO réel)
{
"score": 12,
"verdict": "NO-GO — sous-performance attendue, refaire",
"preset": "ffa",
"metrics": {
"duration_s": 12.6,
"resolution": "1080x1920",
"hook_variance_t2": 1036.1,
"action_density": 6.44,
"iconic_score": 0,
"matchup": ["cristiano_ronaldo", "kylian_mbappe", "lionel_messi", "neymar"]
},
"contributions": {
"duration_s": {"value": 12.6, "verdict": "FAIL", "contribution": 0.0},
"iconic_score": {"value": 0.0, "verdict": "FAIL", "contribution": 0.0},
"action_density": {"value": 6.44, "verdict": "FAIL", "contribution": 0.0},
"hook_variance_t2": {"value": 1036.1, "verdict": "WARN", "contribution": 7.5}
},
"recommendations": [
"FFA trop court (<30s). Re-render impératif.",
"Moins de 2 personnages iconiques. Remplacer par Mario/Pikachu/Sonic/Goku/Steve.",
"FFA statique (< 12). Re-render."
]
}Calibration (24/05/2026)
- Batch 21 FFA : vidéos 13-16s toutes NO-GO (score ≤15)
- Batch 21 FFA réussies (27-29s, icônes gaming) : GO (score ≥70)
- Signal forensic FFA : la durée est le discriminant #1 (delta top vs bottom = 35+ pts)
OUTPUT CONTRACT
Respecte ~/.claude/rules/output-contract.md. Identique à audit-short-1v1 mais seuils FFA.
Attendu : JSON stdout + exit code
Checks spécifiques FFA :
- JSON parseable, champs
score/verdict/preset=ffa/metrics/contributions/recommendations - Seuils FFA appliqués : durée 45-75s PASS, iconic ≥3/4 PASS, action_density ≥18 PASS
- Bonus multi_fighter visible t=2s : +5 si ≥3 fighters distincts détectés
- Si durée < 30s ou > 90s : NO-GO obligatoire
- Si 4 fighters fournis : matchup champ contient 4 items
Format réponse : LIVRABLE: audit FFA / SCORE: X / VERDICT: GO|WARN|NO-GO / DURÉE: Ys (target 45-75) / ICONIC: A/4 / FIGHTERS VIS T2: B / NEXT: <upload | re-render | re-match>
Anti-hallucination : Si seuls 3 fighters détectés sur 4 attendus : flag INCOMPLETE: connected_components < 4 fighters (bug spawn probable).
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

