prompt-translator
Skill28 starsChanged 29 days agoMITView on GitHub
Adapte un prompt d'un modèle/plateforme à un autre en préservant l'intention. À utiliser pour porter un prompt de ChatGPT vers Claude, de Midjourney vers DALL-E, ou entre modèles. Se déclenche aussi avec "adapter pour Claude", "convertir de GPT", "porter ce prompt", "prompt Midjourney", ou toute traduction de prompt entre modèles. Also triggers on "adapt a prompt to another model", "port a prompt", "same prompt for GPT and Claude".
1 other project uses a version of this.
Adopted by
Most recently taken up by
- cclank/lanshu-awesome-ai-video-kittheir own version, first seen Oct 13, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- cclank/lanshu-awesome-ai-video-kit★ 418same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository347 other files it gives its agents.
- agent-a2a-protocol-guideSkill
- agent-agent-observabilitySkill
- agent-api-caller-subagentSkill
- agent-autogen-guideSkill
- agent-browser-agent-builderSkill
- agent-claude-code-extension-builderSkill
- agent-code-review-subagentSkill
- agent-coding-agent-builderSkill
- agent-conflict-resolverSkill
- agent-consensus-builderSkill
- agent-context-managerSkill
- agent-cost-optimizerSkill
- agent-crewai-expertSkill
- agent-customer-support-agentSkill
- agent-data-analyst-agentSkill
- agent-database-query-subagentSkill
- agent-deployment-guideSkill
- agent-email-agent-builderSkill
- agent-evaluation-frameworkSkill
- agent-file-processor-subagentSkill
- agent-handoff-designerSkill
- agent-hierarchy-designerSkill
- agent-human-in-the-loop-designerSkill
- agent-langgraph-designerSkill
- agent-load-balancerSkill
- agent-marketplace-creatorSkill
- agent-mcp-server-builderSkill
- agent-memory-designerSkill
- agent-message-protocolSkill
- agent-monitoring-setupSkill
- agent-multi-agent-orchestratorSkill
- agent-openai-assistants-builderSkill
- agent-pipeline-composerSkill
- agent-pool-managerSkill
- agent-prompt-tunerSkill
- agent-research-agent-designerSkill
- agent-result-aggregatorSkill
- agent-retry-strategistSkill
- agent-sales-agent-builderSkill
- agent-security-hardenerSkill
- agent-semantic-kernel-guideSkill
- agent-skill-creatorSkill
- agent-skill-routerSkill
- agent-spawnerSkill
- agent-state-synchronizerSkill
- agent-subagent-delegatorSkill
- agent-supervisor-builderSkill
- agent-task-decomposerSkill
- agent-testing-frameworkSkill
- agent-tool-calling-architectSkill
- agent-voice-agent-builderSkill
- agent-web-scraper-subagentSkill
- agent-workflow-automation-agentSkill
- ai-ml-anomaly-detection-builderSkill
- ai-ml-computer-vision-guideSkill
- ai-ml-dataset-builderSkill
- ai-ml-ml-experiment-trackerSkill
- ai-ml-model-fine-tunerSkill
- ai-ml-model-optimization-guideSkill
- ai-ml-nlp-pipeline-designerSkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 4 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 157
Never say "happy to help", "sure!", "great question", "let me", or similar.
- Distinctive: line 160
Stop when done. No summary, no recap, no trailing commentary.
- Distinctive: line 161
No politeness wrappers. Direct and blunt.
- Distinctive: line 164
No emoji unless asked.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Prompt Translator
Workflow de traduction en 5 étapes
Étape 1 — Collecte des infos
Demander si absent :
- Modèle source (ex.
gpt-4o,claude-3-5-sonnet,midjourney v6,gemini-1.5-pro) - Modèle cible (ex.
claude-opus-4,dall-e-3,gpt-4.1) - Prompt original complet — system prompt + user message séparés si applicable
- Contexte d'usage (API directe, playground, intégration applicative)
Étape 2 — Diagnostic des différences critiques
Analyser selon la paire source→cible :
| Dimension | GPT-4x | Claude | Gemini | Midjourney/DALL-E |
| Structure préférée | Markdown, instructions linéaires | Balises XML <role>, <context>, <instructions> | Markdown, liste hiérarchique | Paramètres inline (--ar, style tokens) |
| System prompt | {"role":"system","content":"..."} | system= param séparé | system_instruction | N/A |
| Few-shot | Messages alternés user/assistant | <examples> XML ou messages alternés | Turns multimodaux | Tokens de style (ex. in the style of) |
| Longueur optimale | ~2 000 tokens suffit | Supporte 200k, profite des instructions longues | 1M tokens, verbose ok | 77 tokens max (CLIP) |
| Refus/guardrails | Safety layer strict, tournures à éviter | Nuancé, accepte ambiguïté si contexte clair | Modéré | Filtrage image automatique |
Étape 3 — Règles de traduction par cas
GPT → Claude
# Avant (GPT)
system: "You are a helpful assistant. Be concise."
user: "Summarize: {{text}}"
# Après (Claude)
<system>
Vous êtes un assistant expert en synthèse.
<constraints>Répondez en moins de 150 mots. Langue : français.</constraints>
</system>
<task>Résumez le texte suivant :\n<text>{{text}}</text></task>- Remplacer les instructions implicites par des contraintes XML explicites
- Séparer
role,context,task,output_formaten balises distinctes - Supprimer les tournures "You must never…" → préférer "Faites X" (positif)
temperature=0GPT ≈temperature=0Claude (échelles identiques 0–1)
Claude → GPT
# Avant (Claude XML)
<instructions>Liste les 3 points clés. Format JSON.</instructions>
<text>{{text}}</text>
# Après (GPT)
system: "Tu es un assistant analytique. Réponds toujours en JSON valide."
user: "Liste les 3 points clés du texte suivant :\n\n{{text}}\n\nFormat attendu : {\"points\":[...]}"- Aplatir les balises XML en texte structuré Markdown
- Déplacer le format de sortie dans le message user ou en fin de system
- Ajouter
response_format: {"type":"json_object"}si JSON requis (API)
Midjourney → DALL-E 3
# Avant (Midjourney)
/imagine a futuristic city at dusk, cyberpunk, neon lights, rainy streets --ar 16:9 --style raw --v 6.1
# Après (DALL-E 3)
"A wide-angle (16:9 crop) photorealistic render of a futuristic cyberpunk city at dusk.
Neon lights reflect on rain-slicked streets. No text, no watermark.
Style: cinematic, high contrast, volumetric fog."- Traduire
--ar→ description de cadrage ("landscape", "portrait", "square") --style raw→ "photorealistic", "no artistic filter"--v 6→ sans équivalent, préciser "hyper-detailed, current AI render quality"--no X→ "Exclude: X" en fin de prompt DALL-E
Midjourney → Stable Diffusion / Flux
# Avant
/imagine portrait of a samurai, bokeh, 85mm lens --ar 2:3 --style raw
# Après (SD/Flux — positive + negative)
positive: "portrait of a samurai, bokeh background, 85mm lens, photorealistic, sharp focus, 2:3 ratio"
negative: "blurry, cartoon, anime, watermark, low quality, deformed hands"- Toujours générer un prompt négatif explicite
- Ajouter des tokens de qualité :
masterpiece, best quality, 8k - Préciser le sampler si connu :
DPM++ 2M Karras,Euler a
Étape 4 — Livraison du prompt traduit
Fournir :
- Prompt final copier-coller (code block, sans explication mélangée)
- Paramètres API si applicable (temperature, max_tokens, response_format)
Exemple de bloc livrable :
# Claude API
response = client.messages.create(
model="claude-opus-4-5",
max_tokens=1024,
system="<role>Expert en synthèse</role>",
messages=[{"role": "user", "content": "<task>Résumez : {{text}}</task>"}]
)Étape 5 — Notes de migration
Signaler systématiquement :
- Fonctionnalités perdues (ex. vision GPT-4V → pas de vision en mode text-only Claude)
- Fonctionnalités gagnées (ex. contexte 200k Claude vs 128k GPT)
- Comportements différents : ton par défaut, verbosité, format JSON natif
- Paramètres à ajuster :
top_p,temperature, seeds image - Tests recommandés : 3 à 5 inputs représentatifs avant de valider la migration
Garde-fous & anti-patterns
| Anti-pattern | Problème | Correctif |
| Copier-coller le prompt sans adaptation | Format inadapté, résultats dégradés | Toujours appliquer les règles de structure cible |
Traduire --ar par rien | DALL-E ignore le ratio, image carrée | Ajouter le ratio en texte ("landscape 16:9") |
| Garder les balises XML pour GPT | GPT les lit comme texte littéral | Aplatir en Markdown |
| Omettre le prompt négatif pour SD/Flux | Artefacts visuels fréquents | Prompt négatif obligatoire |
| Ignorer les limites de tokens Midjourney | Prompt tronqué silencieusement à 77 tokens | Compresser le prompt, tester le token count |
system vide pour Claude | Comportement générique, pas de persona | Toujours renseigner system= |
| Temperature identique sans vérifier l'échelle | Gemini 0.0–2.0 ≠ Claude 0.0–1.0 | Normaliser selon l'échelle cible |
Paires de traduction courantes (2026)
| Source | Cible | Effort | Notes |
| GPT-4o | Claude Opus/Sonnet | Faible | Structurer en XML, gain sur longs contextes |
| Claude 3.x | GPT-4.1 | Faible | Aplatir XML → Markdown |
| Midjourney v6 | DALL-E 3 | Moyen | Perte des paramètres fins de style |
| Midjourney v6 | Flux / SD 3.5 | Moyen | Nécessite prompt négatif |
| GPT-4o | Gemini 1.5/2.0 | Faible | Adapter system → system_instruction |
| Gemini | Claude | Faible | Ajouter XML, séparer system/user |
Communication Rules — MANDATORY
- Ultra-concise. No filler, no preamble, no pleasantries.
- Never say "happy to help", "sure!", "great question", "let me", or similar.
- Tool first, talk second. Act before explaining.
- Result first. Lead with outcome, not process.
- Stop when done. No summary, no recap, no trailing commentary.
- No politeness wrappers. Direct and blunt.
- Minimum words. If one word works, do not use ten.
- No unsolicited explanations.
- No emoji unless asked.
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.

