diff --git a/skill/SKILL.md b/skill/SKILL.md index b758970..597590c 100644 --- a/skill/SKILL.md +++ b/skill/SKILL.md @@ -2,11 +2,12 @@ name: vibesku description: | VibeSKU CLI skill for generating e-commerce visuals and listing copy from product photos. - Use when users need template-based generation (`ecom-hero`, `kv-image-set`, `exploded-view`, - `image-translation`, `lifestyle-scene`, `white-background`, `listing`), refinement, export/download, batch jobs, auth setup, or - credits/config management. Triggers on requests mentioning VibeSKU workflows, product image - generation, hero banners, exploded views, image/poster translation, white-background packshots, - listing copy, or batch runs. + Use when users need template-based generation (`scenario-set`, `ecom-hero`, `kv-image-set`, + `exploded-view`, `image-translation`, `lifestyle-scene`, `white-background`, `listing`), + refinement, export/download, batch jobs, auth setup, or credits/config management. Triggers on + requests mentioning VibeSKU workflows, product image generation, platform image sets + (Amazon/Taobao/Shopify/RED), hero banners, exploded views, image/poster translation, + white-background packshots, listing copy, or batch runs. metadata: openclaw: requires: @@ -65,10 +66,11 @@ vibesku init vsk_ # API key for CI/CD ## Template Selection Guide -VibeSKU provides 7 templates. **Read the corresponding reference file before building the generate command.** +VibeSKU provides 8 templates. **Read the corresponding reference file before building the generate command.** | Need | Template | Output | Cost | Reference | |------|----------|--------|------|-----------| +| Complete platform image set in one pass (Amazon / Taobao / Shopify / RED / custom) | `scenario-set` | IMAGE | 1-2 cr/img × slots | [scenario-set.md](references/scenario-set.md) | | Single product image (main photo, banner, poster) | `ecom-hero` | IMAGE | 1-2 cr/img | [ecom-hero.md](references/ecom-hero.md) | | Coordinated detail-page poster set | `kv-image-set` | IMAGE | 1-2 cr/img × scenes | [kv-image-set.md](references/kv-image-set.md) | | Single technical exploded infographic | `exploded-view` | IMAGE | 1-2 cr/img | [exploded-view.md](references/exploded-view.md) | @@ -81,6 +83,12 @@ VibeSKU provides 7 templates. **Read the corresponding reference file before bui ``` User wants visuals? +├── Whole listing / platform image set in one pass → scenario-set +│ ├── Amazon listing pack → scenarioId: amazon-pack +│ ├── Taobao/Tmall 主图+详情页 → scenarioId: taobao-pack +│ ├── Shopify storefront pack → scenarioId: shopify-pack +│ ├── RED/Xiaohongshu post set → scenarioId: red-pack +│ └── Hand-picked image types → scenarioId: custom-pack ├── Technical exploded infographic → exploded-view │ ├── Balanced callouts (default) → labelPlacement: balanced-callout │ ├── Cleaner visual without labels → labelPlacement: none diff --git a/skill/VERSION b/skill/VERSION index ef52a64..8f0916f 100644 --- a/skill/VERSION +++ b/skill/VERSION @@ -1 +1 @@ -0.4.6 +0.5.0 diff --git a/skill/references/commands.md b/skill/references/commands.md index c843ca9..7889cb0 100644 --- a/skill/references/commands.md +++ b/skill/references/commands.md @@ -103,6 +103,7 @@ Show detailed template spec: asset requirements, brief fields, options with defa ```bash vibesku templates info ecom-hero # Human-readable vibesku templates info kv-image-set --json # JSON +vibesku templates info scenario-set # Platform image set (Amazon/Taobao/Shopify/RED) template vibesku templates info exploded-view # Exploded infographic template vibesku templates info image-translation # Image/poster text translation template vibesku templates info lifestyle-scene # Realistic product usage scene template diff --git a/skill/references/scenario-set.md b/skill/references/scenario-set.md new file mode 100644 index 0000000..3cb8c18 --- /dev/null +++ b/skill/references/scenario-set.md @@ -0,0 +1,126 @@ +# Template: scenario-set + +**Purpose**: Generate a complete platform-ready image set from one upload in a single pass. Pick a scenario card (a platform × purpose recipe such as an Amazon listing pack) and VibeSKU fans out every selected image slot — white-background packshots, hero images, lifestyle scenes, detail close-ups, infographic posters — as one coordinated batch. + +**When to use**: User is launching or refreshing a COMPLETE listing on a specific platform (Amazon, Taobao/Tmall, Shopify, RED/Xiaohongshu) and wants the whole image set at once instead of running single templates one by one. For one precise image, use the single template (`white-background`, `ecom-hero`, `lifestyle-scene`, etc.) directly. + +**Output type**: IMAGE | **Supports analysis**: Yes | **Cost**: 1K/2K = 1 credit/slot, 4K = 2 credits/slot + +## Assets + +| Role | Min | Max | Required | +|------|-----|-----|----------| +| PRODUCT (product images) | 1 | 10 | Yes | +| LOGO (brand logo) | 0 | 1 | No | + +Upload multiple product angles when possible — some slots (alternate-angle catalog shots) reference a specific photo by upload order and degrade to a best-effort angle change when that photo is missing. + +## Parameters + +| Parameter | Type | Default | Description | +|-----------|------|---------|-------------| +| `scenarioId` | string | — (required) | **Scenario card** — the platform × purpose recipe (see below) | +| `slots` | string[] | — (required) | Slot IDs selected from that scenario's recipe — each slot = 1 output image | +| `imageSize` | string | `2K` | Resolution: `1K`, `2K`, or `4K` | +| `primaryLang` | string | scenario default | Language for on-image text (posters/infographics). Omit or `null` to use the scenario's marketplace default | + +## Scenario Cards + +| Scenario ID | Platform / Purpose | Default language | Slot recipe (slot id → content) | +|-------------|--------------------|------------------|--------------------------------| +| `amazon-pack` | Amazon listing image set | `en` | `main` (white-bg main, locked), `angle-2`/`angle-3` (alt-angle white-bg, best-effort, use photos #2/#3), `lifestyle`, `selling-points`, `detail` | +| `taobao-pack` | Taobao/Tmall 主图 + 详情页 | `zh-Hans` | Main images: `main-1` (hero, locked), `main-2` (alt angle), `main-3` (detail close-up), `main-4` (specs), `main-5` (white-bg). Detail page: `detail-hero` (selling points), `detail-lifestyle` (story poster), `detail-painpoint`, `detail-usage` (off by default) | +| `shopify-pack` | Shopify storefront set | `en` | `banner` (16:9 hero banner, locked), `main` (white-bg), `lifestyle-1`, `lifestyle-2` (in-use), `detail`, `callouts` (selling points) | +| `red-pack` | RED / Xiaohongshu post set | `zh-Hans` | `cover` (3:4 vertical poster, locked), `inner-lifestyle` (in-use), `inner-detail`, `inner-review` (review-style visual) | +| `custom-pack` | Build-your-own set | `en` | One slot per slot type — every slot id equals its slot type id (e.g. `white-bg-main`, `lifestyle`, `exploded-infographic`); pick any combination | + +Notes: + +- **Locked slots** are the anchor image of the pack (e.g. Amazon's white-background main). The dashboard always keeps them on; via API/CLI you may omit them, but including the locked slot is strongly recommended. +- **`slots` must come from the chosen scenario's recipe** — unknown slot ids are ignored, and at least one valid slot is required. +- Slot order in the recipe determines output order. + +## Slot Types (content catalog) + +Each slot maps to one of these content types. `custom-pack` exposes all of them as directly selectable slots. + +| Slot type | Content | Ratio | Renders via | +|-----------|---------|-------|-------------| +| `white-bg-main` | Clean white-background main packshot | 1:1 | white-background engine | +| `white-bg-angle` | Alternate-angle catalog shot (uses matching photo when provided) | 1:1 | white-background engine | +| `hero-main` | Hero main image with styled staging | 1:1 | ecom-hero MAIN_IMAGE | +| `hero-banner` | Wide marketing banner | 16:9 | ecom-hero BANNER | +| `hero-poster` | Vertical promo poster | 3:4 | ecom-hero POSTER | +| `lifestyle` | Product in a realistic context scene | 1:1 | lifestyle-scene | +| `lifestyle-in-use` | Hands-on usage scene | 4:5 | lifestyle-scene (hand-only) | +| `lifestyle-story` | Lifestyle story poster with copy | 3:4 | KV lifestyle scene | +| `selling-points` | Selling-point visual with callouts | 1:1 | KV hero scene | +| `detail-texture` | Material/texture macro close-up | 1:1 | KV detail scene | +| `detail-function` | Functional detail / mechanism shot | 1:1 | KV functional detail scene | +| `specs-attributes` | Specs & attributes table visual | 1:1 | KV specs scene | +| `exploded-infographic` | Technical exploded infographic | 3:4 | exploded-view engine | +| `social-proof` | Review-style trust visual | 3:4 | KV social-proof scene | +| `pain-point` | Pain point vs. solution contrast | 3:4 | KV pain-point scene | +| `brand-story` | Brand story editorial poster | 3:4 | KV brand-story scene | +| `usage-guide` | Step-by-step usage guide | 3:4 | KV usage-guide scene | + +## Product Analysis (supportsAnalysis) + +`scenario-set` jobs support AI photo analysis: the platform reads the uploaded photos with a vision model and pre-fills `productName`, `brandName`, and `productDetails` (selling points), plus per-slot-type presentation ideas that automatically enrich each slot's prompt. In the dashboard this runs automatically after upload. Via API/CLI, provide a complete brief (`-n`, `-d`, `-b`) instead — analysis is not required when the brief is filled in. + +## Examples + +```bash +# Full Amazon listing set (6 images) from three product angles +vibesku generate -t scenario-set \ + -n "Ultimate Omega Softgels" -b "Nordic Naturals" \ + -d "High-absorption fish oil, lemon flavor, 90 softgels" \ + -i front.jpg back.jpg side.jpg \ + -o '{ + "scenarioId": "amazon-pack", + "slots": ["main", "angle-2", "angle-3", "lifestyle", "selling-points", "detail"] + }' + +# Taobao 主图五图 + 详情页三张（默认中文文案） +vibesku generate -t scenario-set \ + -n "智能保温杯" -d "12小时保温，LED 温显" -b "暖芯" \ + -i cup.jpg cup-side.jpg \ + -o '{ + "scenarioId": "taobao-pack", + "slots": ["main-1", "main-2", "main-3", "main-4", "main-5", "detail-hero", "detail-lifestyle", "detail-painpoint"] + }' + +# Shopify storefront set with English copy at 4K +vibesku generate -t scenario-set \ + -n "Minimalist Desk Lamp" -i lamp.jpg \ + -o '{ + "scenarioId": "shopify-pack", + "slots": ["banner", "main", "lifestyle-1", "detail", "callouts"], + "imageSize": "4K" + }' + +# RED post set (cover + 3 inner images) +vibesku generate -t scenario-set \ + -n "氨基酸洗面奶" -i cleanser.jpg \ + -o '{ + "scenarioId": "red-pack", + "slots": ["cover", "inner-lifestyle", "inner-detail", "inner-review"] + }' + +# Custom pick: just a white-bg main, an exploded infographic, and a usage guide +vibesku generate -t scenario-set \ + -n "RGB Gaming Keyboard" -i keyboard.jpg \ + -o '{ + "scenarioId": "custom-pack", + "slots": ["white-bg-main", "exploded-infographic", "usage-guide"], + "primaryLang": "en" + }' +``` + +## Tips + +- **Billing**: Each slot = 1 image = 1 credit (1K/2K) or 2 credits (4K). The Amazon pack at 2K costs 6 credits. +- **Regenerate one slot**: Each output is tagged with its slot id. To redo a single image, run `generate` again with `"slots": [""]` on the same job, or use `vibesku refine ` for an edit instead of a redo. +- **Partial failure**: Slots fan out independently — if one slot fails to start, the rest still run and only successful slots are billed. +- **`primaryLang`** only affects slots that render on-image copy (posters/infographics); packshots and plain scenes are text-free. +- **Choosing between this and `kv-image-set`**: `kv-image-set` produces a stylistically unified vertical poster series for a detail page; `scenario-set` mixes engines (packshots + heroes + scenes + posters) to cover a platform's full image checklist. diff --git a/skill/references/versioning.md b/skill/references/versioning.md index 4d45b7f..6644046 100644 --- a/skill/references/versioning.md +++ b/skill/references/versioning.md @@ -61,5 +61,5 @@ Reference source: After updating, validate with your agent's normal runtime checks: 1. Confirm installed skill metadata shows a newer or expected version. -2. Confirm template docs include `ecom-hero`, `kv-image-set`, `exploded-view`, `image-translation`, `lifestyle-scene`, `white-background`, and `listing`. +2. Confirm template docs include `scenario-set`, `ecom-hero`, `kv-image-set`, `exploded-view`, `image-translation`, `lifestyle-scene`, `white-background`, and `listing`. 3. Re-run the user task that triggered the version check and verify the mismatch/error is resolved.