swarm-assistent/Config/_base/skills/prompting.md

Skill: prompting

The chat model prepares the Generate-box text for Krea 2 (Qwen3-VL). Do this prep as fully as you can in the same turn (or the dedicated prep hop). Chat with the user may be RU/EN; the JSON prompt is always English.

Hard rules (Krea / Qwen3-VL)

1. prompt language = English only — never put Russian into the JSON prompt. Translate the idea, then structure it.
2. Natural photographer/director prose — not Danbooru tag soup, not (word:1.5), not masterpiece / best quality / 8k.
3. Order (front-load):
   subject → pose/action → body/wardrobe → setting → materials/textures → camera/framing → lighting → medium/mood.
4. Put LoRA trigger phrases (exact English spelling) near the subject they affect.
5. negative on every Generate — create if live is empty (Exact generation.negative), supplement if the scene needs a specific omit, or echo live unchanged. Put “no blur / no people” ideas as positives in prompt instead of stuffing the negative box. Never clear negative.
6. Short ideas: expand into a full Krea paragraph. Thin or RU drafts: rewrite before Generate — do not hand Krea a half-ready line.

Prep checklist (before actions:["generate"])

• English only in prompt
• negative present (new / supplemented / echoed — not omitted)
• Subject and action clear in the first sentence
• Wardrobe / body / setting concrete
• Camera + lighting present
• One coherent scene; NSFW stated in plain English if needed
• Triggers placed next to what they modify

Deliverable

User-facing prose: short, in the user’s language.
JSON prompt: English, structured as above — ready for Swarm Generate / Krea 2.

Bad → good

Bad: cute fox, snow, masterpiece, best quality, 8k, detailed, no blur

Good: A fluffy red fox sitting alert in fresh powder snow, ears forward, breath faintly visible in the cold air, soft morning light from the left catching orange fur and casting long blue shadows, shot on an 85mm lens at f/2.8 with creamy bokeh, calm winter atmosphere, sharp eyes and whiskers.