ComfyUI image generation plugin for OpenClaw
- TypeScript 99.5%
- JavaScript 0.5%
cc4962a236
/image-option list models|loras now uses the active workflow's primary
loader enum from the cached schema instead of the global /models/checkpoints
endpoint — only entries the workflow can actually load show up.
Arguments: /image-option list <models|loras> [filter] [page]
- filter: substring match across qualified name + filename (case
insensitive). Matches "(Illustrious)" / "(Pony)" naturally because
qualifiedModelName embeds base_model.
- page: 50 entries per page; defaults to 1.
Each list persists on session.lastListing. /image-option set model|lora <N>
resolves N against that listing (1-indexed), so picking entries with kanji,
emoji, or version-collision names no longer requires typing the name.
LoRAs use the generic LoraLoader.lora_name enum (workflows often inject
the LoraLoader at gen time rather than embedding it statically).
Falls back to legacy live-fetch when no schema cache or loader can be
resolved.
|
||
|---|---|---|
| skills/comfyui | ||
| src | ||
| workflows | ||
| .gitignore | ||
| index.ts | ||
| openclaw.plugin.json | ||
| package-lock.json | ||
| package.json | ||
| test-loader.mjs | ||
| tsconfig.json | ||