Swarm Assistent

SwarmUI extension for collaborative Krea 2 prompting via Ollama: chat + board (Generate | Refs tabs), LoRA chips, personas, model cards with Civitai fetch, img2img/inpaint, slash commands, auto Generate.

Version 0.6.0 — board tabs, Cards form + live Civitai meta, Interrupt cancel, slim history, LoRA chips, Turbo/RAW, taste on disk, RU UI.

Layout

• Left — Board tabs: Generate (full-height live view) | Refs (reference grid + badge N · vision M)
• Splitter: drag to resize panes
• Right: Chat | Cards; persona / pack / Ollama model; settings gear
• Chips: aspect, Seed lock/−1, Vary, Turbo/RAW; LoRA chip row under them
• Slash autocomplete when the composer starts with /

UX

• Send to Assistent under Generate/History → Ref + Assistent tab
• Drop on Generate → new Ref + switch to Refs tab
• As Init / As Mask / Clear Init
• Enter sends; Shift+Enter newline; Interrupt cancels chat epoch (no late auto-apply)
• Manual Apply + Generate / /gen always generate; Auto-generate checkbox only for LLM auto-path
• Auto-critique rewrites prompt only (no second Generate)
• Civitai Confirm required (unless auto-download)
• Cards: form fields + previews; Load Civitai meta fetches by SHA (by-hash); status always explicit

Slash commands (client-side, no LLM)

Command	Effect
/help	List commands
/gen	Generate now
/look generate|refN	Attach that board window + ask the LLM to look
/init /mask /clear	Same as board buttons
/interrupt	Stop generation / cancel chat
/aspect 16:9	Set size from the official 1K table
/seed lock|random	Lock or randomize seed
/vary	New seed, same prompt (+ generate if auto)
/pack write|critique|compose|params|inpaint|describe|card	Switch pack
/civitai <query>	Ask LLM to search Civitai (Confirm still required)
/inventory	Rescan models + refresh LoRA list

Requirements

• SwarmUI with a Krea 2 checkpoint selected
• Ollama on http://127.0.0.1:11434 on the GPU VM (gpu-rent LLM_RUNTIME=ollama)
• At least one pulled model
• Optional: Civitai API key in SwarmUI User Settings

Install

swarmui:
  - url: https://gitea.hsrv.site/mrleo1nid/swarm-assistent.git
    ref: main
    dir: swarm-assistent
    requires: ollama

Restart / rebuild SwarmUI after clone. gpu-rent: seed-extensions + restart.

Prompt packs

Pack	Role
base_krea2	Always injected
write_prompt	Craft / improve prompts
critique_image	Vision critique
compose_scene	Scene via board refs
fix_params	Aspect / steps / CFG / seed (krea_profile in live context)
inpaint_edit	Init + Mask
describe_ref	Vision → prompt
catalog_card	Recommendation card JSON

Personas: neutral, lewd, aggressive, cinema, terse. Overlay: /mnt/swarm_data/Assistent/personas.json (gpu-rent seeds from assistent-personas.yaml → yaml + json). Assistent reads json only.

Taste: /mnt/swarm_data/Assistent/taste.json via AssistentGetTaste / AssistentSaveTaste (merged with browser localStorage by updated).

Cards: {stem}.assistent.json next to weights; Civitai sidecar {stem}.civitai.json. Wanted queue → .gpu-rent-wanted-models.yaml.

API routes

Route	Role
AssistentListModels	Ollama /api/tags
AssistentListInventory	LoRA / checkpoint / wildcard inventory
AssistentListPersonas	Bundled + overlay personas
AssistentGetPacks	Prompt pack texts
AssistentGetCard / AssistentSaveCard	.assistent.json cards
AssistentGetCardMeta	Local sidecar + optional fetch=true Civitai by-hash
AssistentEnqueueWanted	Wanted YAML queue
AssistentGetTaste / AssistentSaveTaste	Persistent taste profile
AssistentSearchCivitai	Civitai LoRA search
AssistentChat	HTTP chat (+ Civitai hop)
AssistentChatWS	Streaming chat WebSocket

License

MIT