@mcptoolshop/style-dataset-lab 1.1.0

package/README.ja.md

@@ -0,0 +1,142 @@
+<p align="center">
+  <a href="README.md">English</a> | <a href="README.zh.md">中文</a> | <a href="README.es.md">Español</a> | <a href="README.fr.md">Français</a> | <a href="README.hi.md">हिन्दी</a> | <a href="README.it.md">Italiano</a> | <a href="README.pt-BR.md">Português (BR)</a>
+</p>
+
+<p align="center">
+  <img src="https://raw.githubusercontent.com/mcp-tool-shop-org/brand/main/logos/style-dataset-lab/readme.png" width="400">
+</p>
+
+<p align="center">
+  <a href="https://www.npmjs.com/package/@mcptoolshop/style-dataset-lab"><img src="https://img.shields.io/npm/v/@mcptoolshop/style-dataset-lab" alt="npm"></a>
+  <a href="LICENSE"><img src="https://img.shields.io/badge/license-MIT-blue" alt="MIT License"></a>
+</p>
+
+# style-dataset-lab
+
+Visual dataset factory：VLMのファインチューニングのための、多様なトレーニングデータを生成、キュレーション、エクスポートします。
+
+## これは何ですか？
+
+「学習可能なビジュアル真実」を構築するためのツールキットです。各アセットには、以下の3つの情報が含まれています。
+
+1. **画像ピクセル**：ComfyUIによって生成され、完全な情報（チェックポイント、LoRA、シード値、サンプラー、CFG）が付与されています。
+2. **規範的な説明**：なぜこの画像が特定のスタイルに合致しているか、または合致していないかを、スタイルに関する規定に基づいて説明します。
+3. **品質評価**：各次元のスコアと参照ルールに基づいて、承認/却下の判断がなされます。
+
+生成されたデータは、[`@mcptoolshop/repo-dataset`](https://github.com/mcp-tool-shop-org/repo-dataset)に連携し、10種類の形式（TRL、LLaVA、Qwen2-VL、Axolotl、LLaMA-Factory、ShareGPT、RLHFペア、DPO、キャプション、分類）で、多様なトレーニングデータを作成します。
+
+## セキュリティモデル
+
+**ローカル環境のみ**。style-dataset-labは、`localhost:8188`で動作するComfyUIとのみ通信し、外部ネットワークへのアクセスは一切行いません。テレメトリー、分析、データ送信機能は一切ありません。画像生成はすべて、お客様のGPU上で行われます。記録データと規範データは、お客様のファイルシステム上に保存されます。
+
+## データセットの統計
+
+| 指標 | 値 |
+|--------|-------|
+| キュレーションされたデータ | 1,182 |
+| 総アセット数 | 2,571件（承認済み：893件、ペインタリーバリアント：887件） |
+| プロンプトの種類 | 28 |
+| ビジュアルカテゴリ | 18種類（衣装、船、内装、装備、環境、生物、ステーション、標識、照明、貨物、アーキテクチャ、生物、表面、日常生活、惑星、損傷/修理、異星生物、生活感のあるディテール） |
+| ペア比較 | 人間による評価：6件、合成データ：71件 |
+| 拒否の種類 | 30+ |
+| IDパケットシステム | 派閥、ランク、視覚的な特徴を持つ名前付きキャラクター |
+| エクスポート形式 | 10種類（@mcptoolshop/repo-dataset経由） |
+
+## インストール
+
+```bash
+npm install -g @mcptoolshop/style-dataset-lab
+```
+
+次に、プロジェクトをクローンするか、新しいデータセットの作業領域を初期化します。
+
+```bash
+git clone https://github.com/mcp-tool-shop-org/style-dataset-lab my-dataset
+cd my-dataset
+npm install
+```
+
+## ワークフロー
+
+```bash
+# 1. Start ComfyUI
+# (point it at your checkpoint + LoRA setup)
+
+# 2. Generate candidates from a prompt pack
+npm run generate -- inputs/prompts/wave1.json
+npm run generate -- inputs/prompts/wave1.json --dry-run
+
+# 3. Generate identity-packet characters
+npm run generate:identity -- inputs/identity-packets/wave27a-identity-spine.json
+
+# 4. Generate painterly variants of approved assets
+npm run painterly -- <asset_id>
+
+# 5. Curate -- approve, reject, or mark borderline
+npm run curate -- <asset_id> approved "explanation" --scores "silhouette:0.9,palette:0.8"
+npm run curate -- <asset_id> rejected "explanation" --failures "too_clean,wrong_material"
+
+# 6. Bind canon explanations to assets
+npm run canon-bind -- <asset_id>
+
+# 7. Record pairwise comparisons
+npm run compare -- <asset_a> <asset_b> a "A has better faction read because..."
+
+# 8. Export training data via repo-dataset
+npm run export
+npm run inspect
+npm run validate
+```
+
+## ディレクトリ構造
+
+```
+canon/                  Style constitution, review rubric, identity gates, species canon
+inputs/
+  prompts/              Prompt packs per wave (JSON: subjects, variations, defaults)
+  references/           IP-Adapter reference images
+  control/              ControlNet control images
+  control-guides/       ControlNet guide overlays
+  identity-packets/     Named character identity spines (faction DNA, rank, visual anchors)
+outputs/
+  candidates/           Raw generations (gitignored)
+  approved/             Curated approved (gitignored)
+  rejected/             Curated rejected (gitignored)
+  borderline/           Curated borderline (gitignored)
+  painterly/            Painterly-style variants (gitignored)
+records/                Per-asset JSON (provenance + judgment + canon binding)
+comparisons/            A-vs-B preference judgments
+exports/                repo-dataset output (gitignored)
+scripts/                generate, curate, compare, canon-bind, painterly, identity gen
+workflows/              Reusable ComfyUI workflow templates
+```
+
+## 生成設定
+
+```yaml
+checkpoint: dreamshaperXL_v21TurboDPMSDE.safetensors
+lora: classipeintxl_v21.safetensors (weight: 1.0)
+resolution: 1024x1024
+steps: 8
+cfg: 2.0
+sampler: dpmpp_sde
+scheduler: karras
+speed: ~9s per image (RTX 5080)
+```
+
+追加の生成モード：ControlNet（ポーズ/深度ガイド）、IP-Adapter（参照ベース）、IDパケット（名前付きキャラクターの一貫性）。
+
+## 必要なもの
+
+- `localhost:8188`で動作する[ComfyUI](https://github.com/comfyanonymous/ComfyUI)
+- DreamShaper XL Turbo チェックポイント + ClassipeintXL LoRA
+- Node.js 20以上
+- トレーニング用エクスポートのための[`@mcptoolshop/repo-dataset`](https://github.com/mcp-tool-shop-org/repo-dataset)
+
+## ライセンス
+
+MIT
+
+---
+
+開発：<a href="https://mcp-tool-shop.github.io/">MCP Tool Shop</a>

package/README.md

@@ -0,0 +1,178 @@
+<p align="center">
+  <a href="README.ja.md">日本語</a> | <a href="README.zh.md">中文</a> | <a href="README.es.md">Español</a> | <a href="README.fr.md">Français</a> | <a href="README.hi.md">हिन्दी</a> | <a href="README.it.md">Italiano</a> | <a href="README.pt-BR.md">Português (BR)</a>
+</p>
+
+<p align="center">
+  <img src="https://raw.githubusercontent.com/mcp-tool-shop-org/brand/main/logos/style-dataset-lab/readme.png" width="400">
+</p>
+
+<p align="center">
+  <a href="https://www.npmjs.com/package/@mcptoolshop/style-dataset-lab"><img src="https://img.shields.io/npm/v/@mcptoolshop/style-dataset-lab" alt="npm"></a>
+  <a href="LICENSE"><img src="https://img.shields.io/badge/license-MIT-blue" alt="MIT License"></a>
+</p>
+
+# style-dataset-lab
+
+Visual dataset factory -- generate, curate, and export multimodal training data for VLM fine-tuning.
+
+## What this is
+
+A **monorepo** toolkit for building **trainable visual truth** across multiple games. Each game gets its own data directory; scripts are shared. Every asset carries three things:
+
+1. **Image pixels** -- generated by ComfyUI with full provenance (checkpoint, LoRA, seed, sampler, cfg)
+2. **Canon explanation** -- why this is on-style or off-style, grounded in a style constitution
+3. **Quality judgment** -- approved/rejected with per-dimension scores and cited rules
+
+The output feeds into [`@mcptoolshop/repo-dataset`](https://github.com/mcp-tool-shop-org/repo-dataset) to produce multimodal training data in 10 formats: TRL, LLaVA, Qwen2-VL, Axolotl, LLaMA-Factory, ShareGPT, RLHF pairs, DPO, captioning, and classification.
+
+## Security model
+
+**Local-only.** style-dataset-lab talks to ComfyUI on `localhost:8188` and never makes external network requests. No telemetry, no analytics, no phoning home. Image generation happens entirely on your GPU. Records and canon data stay on your filesystem.
+
+## Dataset stats (Star Freight)
+
+The included `games/star-freight/` dataset ships with the repo as a working example:
+
+| Metric | Value |
+|--------|-------|
+| Curated records | 1,182 |
+| Total assets | 2,571 (893 approved, 887 painterly variants) |
+| Prompt waves | 28 |
+| Visual categories | 18 (costumes, ships, interiors, equipment, environments, species, stations, signage, lighting, cargo, architecture, creatures, surfaces, daily life, planetary, damage/repair, alien biology, lived-in details) |
+| Pairwise comparisons | 6 human + 71 synthetic |
+| Distinct rejection types | 30+ |
+| Identity packet system | Named characters with faction DNA, rank, and visual anchors |
+| Export formats | 10 (via @mcptoolshop/repo-dataset) |
+
+## Install
+
+```bash
+# Install scripts globally (optional)
+npm install -g @mcptoolshop/style-dataset-lab
+
+# Clone the repo (includes star-freight game data)
+git clone https://github.com/mcp-tool-shop-org/style-dataset-lab
+cd style-dataset-lab
+npm install
+```
+
+To start a new game, create its directory structure:
+
+```bash
+mkdir -p games/my-game/{canon,records,comparisons,inputs/prompts,outputs/{candidates,approved,rejected},exports}
+```
+
+Or use the existing `games/star-freight/` data to explore the workflow immediately.
+
+## Monorepo structure
+
+Each game lives in `games/<name>/` with its own canon, records, and assets. Scripts are shared and accept `--game <name>` (defaults to `star-freight`).
+
+```
+style-dataset-lab/
+  scripts/                  Shared tooling (generate, curate, compare, etc.)
+  games/
+    star-freight/           Star Freight universe data
+      canon/                Style constitution, review rubric, species canon
+      records/              Per-asset JSON (provenance + judgment + canon)
+      comparisons/          A-vs-B preference judgments
+      inputs/               Prompt packs, identity packets, references
+      outputs/              Generated images (gitignored)
+      exports/              repo-dataset output (gitignored)
+    fractured-road/         (add your next game here)
+      canon/
+      records/
+      ...
+```
+
+## Workflow
+
+```bash
+# 1. Start ComfyUI
+# (point it at your checkpoint + LoRA setup)
+
+# 2. Generate candidates from a prompt pack
+npm run generate -- --game star-freight inputs/prompts/wave1.json
+npm run generate -- --game star-freight inputs/prompts/wave1.json --dry-run
+
+# 3. Generate identity-packet characters
+npm run generate:identity -- --game star-freight inputs/identity-packets/wave27a.json
+
+# 4. Generate painterly variants of approved assets
+npm run painterly -- --game star-freight
+
+# 5. Curate -- approve, reject, or mark borderline
+npm run curate -- --game star-freight <asset_id> approved "explanation"
+npm run curate -- --game star-freight <asset_id> rejected "explanation" --failures "too_clean"
+
+# 6. Bind canon explanations to assets
+npm run canon-bind -- --game star-freight
+
+# 7. Record pairwise comparisons
+npm run compare -- --game star-freight <asset_a> <asset_b> a "A has better faction read"
+
+# 8. Export training data via repo-dataset
+repo-dataset visual generate ./games/star-freight --format trl
+repo-dataset visual inspect ./games/star-freight
+```
+
+## Adding a new game
+
+```bash
+mkdir -p games/my-game/{canon,records,comparisons,inputs/prompts,outputs/{candidates,approved,rejected},exports}
+# Write your canon/constitution.md and canon/review-rubric.md
+# Create prompt packs in inputs/prompts/
+# Then generate, curate, and export as above with --game my-game
+```
+
+## Per-game directory layout
+
+Each `games/<name>/` directory contains:
+
+```
+canon/                  Style constitution, review rubric, species canon, identity gates
+inputs/
+  prompts/              Prompt packs per wave (JSON: subjects, variations, defaults)
+  references/           IP-Adapter reference images
+  control-guides/       ControlNet guide overlays
+  identity-packets/     Named character identity spines
+outputs/
+  candidates/           Raw generations (gitignored)
+  approved/             Curated approved (gitignored)
+  rejected/             Curated rejected (gitignored)
+  borderline/           Curated borderline (gitignored)
+  painterly/            Painterly-style variants (gitignored)
+records/                Per-asset JSON (provenance + judgment + canon binding)
+comparisons/            A-vs-B preference judgments
+exports/                repo-dataset output (gitignored)
+```
+
+## Generation setup
+
+```yaml
+checkpoint: dreamshaperXL_v21TurboDPMSDE.safetensors
+lora: classipeintxl_v21.safetensors (weight: 1.0)
+resolution: 1024x1024
+steps: 8
+cfg: 2.0
+sampler: dpmpp_sde
+scheduler: karras
+speed: ~9s per image (RTX 5080)
+```
+
+Additional generation modes: ControlNet (pose/depth-guided), IP-Adapter (reference-driven), and identity packets (named character consistency).
+
+## Requirements
+
+- [ComfyUI](https://github.com/comfyanonymous/ComfyUI) running on localhost:8188
+- DreamShaper XL Turbo checkpoint + ClassipeintXL LoRA
+- Node.js 20+
+- [`@mcptoolshop/repo-dataset`](https://github.com/mcp-tool-shop-org/repo-dataset) for training export
+
+## License
+
+MIT
+
+---
+
+Built by <a href="https://mcp-tool-shop.github.io/">MCP Tool Shop</a>

package/README.pt-BR.md

@@ -0,0 +1,142 @@
+<p align="center">
+  <a href="README.ja.md">日本語</a> | <a href="README.zh.md">中文</a> | <a href="README.es.md">Español</a> | <a href="README.fr.md">Français</a> | <a href="README.hi.md">हिन्दी</a> | <a href="README.it.md">Italiano</a> | <a href="README.md">English</a>
+</p>
+
+<p align="center">
+  <img src="https://raw.githubusercontent.com/mcp-tool-shop-org/brand/main/logos/style-dataset-lab/readme.png" width="400">
+</p>
+
+<p align="center">
+  <a href="https://www.npmjs.com/package/@mcptoolshop/style-dataset-lab"><img src="https://img.shields.io/npm/v/@mcptoolshop/style-dataset-lab" alt="npm"></a>
+  <a href="LICENSE"><img src="https://img.shields.io/badge/license-MIT-blue" alt="MIT License"></a>
+</p>
+
+# style-dataset-lab
+
+Ferramenta para criação de conjuntos de dados visuais – gera, organiza e exporta dados de treinamento multimodais para o ajuste fino de modelos de linguagem visuais (VLM).
+
+## O que é isso
+
+Um conjunto de ferramentas para construir a "verdade visual" que pode ser usada para treinamento. Cada elemento contém três coisas:
+
+1. **Pixels da imagem** – gerados pelo ComfyUI com total rastreabilidade (checkpoint, LoRA, seed, sampler, cfg).
+2. **Explicação detalhada** – por que este elemento está ou não de acordo com o estilo definido, com base em uma "constituição" de estilo.
+3. **Avaliação de qualidade** – aprovado/rejeitado, com pontuações por dimensão e regras citadas.
+
+A saída é integrada ao [`@mcptoolshop/repo-dataset`](https://github.com/mcp-tool-shop-org/repo-dataset) para produzir dados de treinamento multimodais em 10 formatos: TRL, LLaVA, Qwen2-VL, Axolotl, LLaMA-Factory, ShareGPT, pares para RLHF, DPO, legendagem e classificação.
+
+## Modelo de segurança
+
+**Apenas local.** O style-dataset-lab se comunica com o ComfyUI em `localhost:8188` e nunca faz solicitações de rede externas. Sem telemetria, sem análise, sem envio de dados. A geração de imagens ocorre inteiramente na sua GPU. Os registros e os dados de referência permanecem no seu sistema de arquivos.
+
+## Estatísticas do conjunto de dados
+
+| Métrica | Valor |
+|--------|-------|
+| Registros organizados | 1,182 |
+| Total de elementos | 2.571 (893 aprovados, 887 variações com estilo pictórico) |
+| Variações de prompts | 28 |
+| Categorias visuais | 18 (roupas, navios, interiores, equipamentos, ambientes, espécies, estações, sinalização, iluminação, carga, arquitetura, criaturas, superfícies, vida cotidiana, planetas, danos/reparos, biologia alienígena, detalhes realistas) |
+| Comparativos | 6 criados por humanos + 71 criados sinteticamente |
+| Tipos de rejeição distintos | 30+ |
+| Sistema de pacotes de identidade | Personagens nomeados com DNA de facção, patente e referências visuais. |
+| Formatos de exportação | 10 (via @mcptoolshop/repo-dataset) |
+
+## Instalação
+
+```bash
+npm install -g @mcptoolshop/style-dataset-lab
+```
+
+Em seguida, clone um projeto ou inicialize um novo espaço de trabalho para conjuntos de dados:
+
+```bash
+git clone https://github.com/mcp-tool-shop-org/style-dataset-lab my-dataset
+cd my-dataset
+npm install
+```
+
+## Fluxo de trabalho
+
+```bash
+# 1. Start ComfyUI
+# (point it at your checkpoint + LoRA setup)
+
+# 2. Generate candidates from a prompt pack
+npm run generate -- inputs/prompts/wave1.json
+npm run generate -- inputs/prompts/wave1.json --dry-run
+
+# 3. Generate identity-packet characters
+npm run generate:identity -- inputs/identity-packets/wave27a-identity-spine.json
+
+# 4. Generate painterly variants of approved assets
+npm run painterly -- <asset_id>
+
+# 5. Curate -- approve, reject, or mark borderline
+npm run curate -- <asset_id> approved "explanation" --scores "silhouette:0.9,palette:0.8"
+npm run curate -- <asset_id> rejected "explanation" --failures "too_clean,wrong_material"
+
+# 6. Bind canon explanations to assets
+npm run canon-bind -- <asset_id>
+
+# 7. Record pairwise comparisons
+npm run compare -- <asset_a> <asset_b> a "A has better faction read because..."
+
+# 8. Export training data via repo-dataset
+npm run export
+npm run inspect
+npm run validate
+```
+
+## Estrutura de diretórios
+
+```
+canon/                  Style constitution, review rubric, identity gates, species canon
+inputs/
+  prompts/              Prompt packs per wave (JSON: subjects, variations, defaults)
+  references/           IP-Adapter reference images
+  control/              ControlNet control images
+  control-guides/       ControlNet guide overlays
+  identity-packets/     Named character identity spines (faction DNA, rank, visual anchors)
+outputs/
+  candidates/           Raw generations (gitignored)
+  approved/             Curated approved (gitignored)
+  rejected/             Curated rejected (gitignored)
+  borderline/           Curated borderline (gitignored)
+  painterly/            Painterly-style variants (gitignored)
+records/                Per-asset JSON (provenance + judgment + canon binding)
+comparisons/            A-vs-B preference judgments
+exports/                repo-dataset output (gitignored)
+scripts/                generate, curate, compare, canon-bind, painterly, identity gen
+workflows/              Reusable ComfyUI workflow templates
+```
+
+## Configuração de geração
+
+```yaml
+checkpoint: dreamshaperXL_v21TurboDPMSDE.safetensors
+lora: classipeintxl_v21.safetensors (weight: 1.0)
+resolution: 1024x1024
+steps: 8
+cfg: 2.0
+sampler: dpmpp_sde
+scheduler: karras
+speed: ~9s per image (RTX 5080)
+```
+
+Modos de geração adicionais: ControlNet (guiado por pose/profundidade), IP-Adapter (guiado por referência) e pacotes de identidade (consistência de personagens nomeados).
+
+## Requisitos
+
+- [ComfyUI](https://github.com/comfyanonymous/ComfyUI) em execução em localhost:8188
+- Checkpoint DreamShaper XL Turbo + LoRA ClassipeintXL
+- Node.js 20+
+- [`@mcptoolshop/repo-dataset`](https://github.com/mcp-tool-shop-org/repo-dataset) para exportação de treinamento
+
+## Licença
+
+MIT
+
+---
+
+Criado por <a href="https://mcp-tool-shop.github.io/">MCP Tool Shop</a

package/README.zh.md

@@ -0,0 +1,142 @@
+<p align="center">
+  <a href="README.ja.md">日本語</a> | <a href="README.md">English</a> | <a href="README.es.md">Español</a> | <a href="README.fr.md">Français</a> | <a href="README.hi.md">हिन्दी</a> | <a href="README.it.md">Italiano</a> | <a href="README.pt-BR.md">Português (BR)</a>
+</p>
+
+<p align="center">
+  <img src="https://raw.githubusercontent.com/mcp-tool-shop-org/brand/main/logos/style-dataset-lab/readme.png" width="400">
+</p>
+
+<p align="center">
+  <a href="https://www.npmjs.com/package/@mcptoolshop/style-dataset-lab"><img src="https://img.shields.io/npm/v/@mcptoolshop/style-dataset-lab" alt="npm"></a>
+  <a href="LICENSE"><img src="https://img.shields.io/badge/license-MIT-blue" alt="MIT License"></a>
+</p>
+
+# style-dataset-lab
+
+视觉数据集工厂 -- 用于生成、整理和导出多模态训练数据，以进行大型视觉语言模型（VLM）的微调。
+
+## 这是什么
+
+一个用于**构建可训练的视觉真理**的工具集。每个资源都包含以下三方面：
+
+1. **图像像素** -- 由 ComfyUI 生成，并包含完整的溯源信息（检查点、LoRA、种子、采样器、CFG）。
+2. **规范解释** -- 解释为什么该图像符合或不符合特定风格，并基于风格规范进行判断。
+3. **质量评估** -- 经过批准/拒绝，并提供每个维度的评分以及引用规则。
+
+输出结果通过 [`@mcptoolshop/repo-dataset`](https://github.com/mcp-tool-shop-org/repo-dataset) 生成多模态训练数据，提供 10 种格式：TRL、LLaVA、Qwen2-VL、Axolotl、LLaMA-Factory、ShareGPT、RLHF 对、DPO、图像描述和分类。
+
+## 安全模型
+
+**仅本地运行。** style-dataset-lab 只与本地 `localhost:8188` 上的 ComfyUI 进行通信，不会进行任何外部网络请求。没有遥测数据，没有分析功能，不会向外部发送数据。图像生成完全在您的 GPU 上进行。记录和规范数据都存储在您的文件系统中。
+
+## 数据集统计信息
+
+| 指标 | 数值 |
+|--------|-------|
+| 整理记录 | 1,182 |
+| 总资源数 | 2,571 (其中 893 个已批准，887 个为绘画风格变体) |
+| 提示词组合 | 28 |
+| 视觉类别 | 18 种 (包括：服装、船只、室内、设备、环境、物种、站点、标识、照明、货物、**架构**、生物、表面、日常生活、行星、损坏/修复、外星生物、细节)。 |
+| 成对比较 | 6 个人工标注 + 71 个合成数据 |
+| 不同类型的拒绝 | 30+ |
+| 身份包系统 | 具有派系 DNA、等级和视觉特征的命名角色。 |
+| 导出格式 | 10 种 (通过 `@mcptoolshop/repo-dataset`) |
+
+## 安装
+
+```bash
+npm install -g @mcptoolshop/style-dataset-lab
+```
+
+然后克隆一个项目或初始化一个新的数据集工作区：
+
+```bash
+git clone https://github.com/mcp-tool-shop-org/style-dataset-lab my-dataset
+cd my-dataset
+npm install
+```
+
+## 工作流程
+
+```bash
+# 1. Start ComfyUI
+# (point it at your checkpoint + LoRA setup)
+
+# 2. Generate candidates from a prompt pack
+npm run generate -- inputs/prompts/wave1.json
+npm run generate -- inputs/prompts/wave1.json --dry-run
+
+# 3. Generate identity-packet characters
+npm run generate:identity -- inputs/identity-packets/wave27a-identity-spine.json
+
+# 4. Generate painterly variants of approved assets
+npm run painterly -- <asset_id>
+
+# 5. Curate -- approve, reject, or mark borderline
+npm run curate -- <asset_id> approved "explanation" --scores "silhouette:0.9,palette:0.8"
+npm run curate -- <asset_id> rejected "explanation" --failures "too_clean,wrong_material"
+
+# 6. Bind canon explanations to assets
+npm run canon-bind -- <asset_id>
+
+# 7. Record pairwise comparisons
+npm run compare -- <asset_a> <asset_b> a "A has better faction read because..."
+
+# 8. Export training data via repo-dataset
+npm run export
+npm run inspect
+npm run validate
+```
+
+## 目录结构
+
+```
+canon/                  Style constitution, review rubric, identity gates, species canon
+inputs/
+  prompts/              Prompt packs per wave (JSON: subjects, variations, defaults)
+  references/           IP-Adapter reference images
+  control/              ControlNet control images
+  control-guides/       ControlNet guide overlays
+  identity-packets/     Named character identity spines (faction DNA, rank, visual anchors)
+outputs/
+  candidates/           Raw generations (gitignored)
+  approved/             Curated approved (gitignored)
+  rejected/             Curated rejected (gitignored)
+  borderline/           Curated borderline (gitignored)
+  painterly/            Painterly-style variants (gitignored)
+records/                Per-asset JSON (provenance + judgment + canon binding)
+comparisons/            A-vs-B preference judgments
+exports/                repo-dataset output (gitignored)
+scripts/                generate, curate, compare, canon-bind, painterly, identity gen
+workflows/              Reusable ComfyUI workflow templates
+```
+
+## 生成设置
+
+```yaml
+checkpoint: dreamshaperXL_v21TurboDPMSDE.safetensors
+lora: classipeintxl_v21.safetensors (weight: 1.0)
+resolution: 1024x1024
+steps: 8
+cfg: 2.0
+sampler: dpmpp_sde
+scheduler: karras
+speed: ~9s per image (RTX 5080)
+```
+
+其他生成模式：ControlNet（姿势/深度引导）、IP-Adapter（参考驱动）以及身份包（命名角色一致性）。
+
+## 要求
+
+- 运行在 `localhost:8188` 上的 [ComfyUI](https://github.com/comfyanonymous/ComfyUI)
+- DreamShaper XL Turbo 检查点 + ClassipeintXL LoRA
+- Node.js 20+
+- [`@mcptoolshop/repo-dataset`](https://github.com/mcp-tool-shop-org/repo-dataset) 用于训练数据导出
+
+## 许可证
+
+MIT
+
+---
+
+由 <a href="https://mcp-tool-shop.github.io/">MCP Tool Shop</a> 构建。

package/SECURITY.md

@@ -0,0 +1,28 @@
+# Security Policy
+
+## Supported Versions
+
+| Version | Supported |
+|---------|-----------|
+| 0.x     | Yes       |
+
+## Threat Model
+
+style-dataset-lab is a **local-only** visual dataset factory. It connects to a ComfyUI server running on `localhost:8188` and does not make any external network calls.
+
+**Attack surface:**
+
+- **ComfyUI server** listens on localhost only. If exposed to a network, an attacker could inject workflow payloads. Keep the default localhost binding.
+- **Untrusted model weights** (`.safetensors`, `.ckpt`, `.pth`, `.bin`) loaded by ComfyUI could contain malicious pickle payloads. Only load weights from sources you trust.
+- **No telemetry.** This tool collects no usage data and makes no outbound requests beyond the local ComfyUI API.
+- **No authentication.** The local ComfyUI API has no auth layer. This is acceptable for localhost but dangerous if network-exposed.
+
+## Reporting a Vulnerability
+
+If you discover a security issue, please email:
+
+**64996768+mcp-tool-shop@users.noreply.github.com**
+
+Expected response time: **72 hours**.
+
+Please do not open public issues for security vulnerabilities.

package/package.json

@@ -0,0 +1,45 @@
+{
+  "name": "@mcptoolshop/style-dataset-lab",
+  "version": "1.1.0",
+  "description": "Visual dataset factory — generate, curate, and export multimodal training data for VLM fine-tuning",
+  "type": "module",
+  "scripts": {
+    "generate": "node scripts/generate.js",
+    "curate": "node scripts/curate.js",
+    "compare": "node scripts/compare.js",
+    "canon-bind": "node scripts/canon-bind.js",
+    "painterly": "node scripts/painterly.js",
+    "generate:identity": "node scripts/generate-identity.js",
+    "export": "echo 'Usage: repo-dataset visual generate ./games/<game> --format trl'",
+    "inspect": "echo 'Usage: repo-dataset visual inspect ./games/<game>'"
+  },
+  "files": [
+    "scripts/",
+    "games/*/canon/",
+    "games/*/inputs/",
+    "package.json",
+    "README.md",
+    "LICENSE",
+    "SECURITY.md",
+    "CHANGELOG.md"
+  ],
+  "keywords": [
+    "dataset",
+    "visual",
+    "training-data",
+    "comfyui",
+    "vlm",
+    "fine-tuning",
+    "multimodal"
+  ],
+  "homepage": "https://mcp-tool-shop-org.github.io/style-dataset-lab/",
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/mcp-tool-shop-org/style-dataset-lab.git"
+  },
+  "engines": {
+    "node": ">=20"
+  },
+  "author": "mcp-tool-shop",
+  "license": "MIT"
+}