meigen 0.1.0

package/README.md

@@ -0,0 +1,242 @@
+<p align="center">
+  <img src="assets/banner.jpg" alt="LibreArt Banner" width="600">
+</p>
+
+<h1 align="center">
+  LibreArt — Visual Creative Expert
+</h1>
+
+<p align="center">
+  <strong>Turn your AI assistant into a creative director for image generation</strong><br>
+  <sub>MCP plugin for Claude Code, OpenClaw, and any MCP-compatible host</sub>
+</p>
+
+<p align="center">
+  <a href="https://www.meigen.ai"><img src="https://img.shields.io/badge/Gallery-meigen.ai-blue?style=flat-square" alt="Gallery"></a>
+  <a href="#tools"><img src="https://img.shields.io/badge/Tools-7-green?style=flat-square" alt="7 Tools"></a>
+  <a href="#providers"><img src="https://img.shields.io/badge/Providers-3-orange?style=flat-square" alt="3 Providers"></a>
+  <a href="LICENSE"><img src="https://img.shields.io/badge/License-MIT-lightgrey?style=flat-square" alt="MIT"></a>
+</p>
+
+<p align="center">
+  <a href="#quick-start">Quick Start</a> •
+  <a href="#what-can-it-do">What Can It Do</a> •
+  <a href="#providers">Providers</a> •
+  <a href="#configuration">Configuration</a>
+</p>
+
+<p align="center">
+  <strong>English</strong> | <a href="README.zh-CN.md">中文</a>
+</p>
+
+---
+
+## Why LibreArt?
+
+Most AI image tools are either simple prompt-in-image-out APIs, or complex UIs that require expertise.
+
+LibreArt takes a different approach: it gives your AI assistant **professional creative knowledge** — 1,300+ curated trending prompts, style-specific enhancement techniques, and multi-step workflow orchestration — so you can describe what you want in plain language, and get production-quality results.
+
+**No prompt engineering skills needed.** Just talk to your AI assistant like you would to a creative director.
+
+---
+
+## Installation
+
+### Claude Code
+
+One command — no clone, no build:
+
+```bash
+claude mcp add --transport stdio libreart -- npx -y meigen
+```
+
+To pass an API key at install time:
+
+```bash
+claude mcp add --transport stdio libreart \
+  --env MEIGEN_API_TOKEN=meigen_sk_xxx \
+  -- npx -y meigen
+```
+
+Or configure later via the interactive setup wizard:
+
+```
+/meigen:setup
+```
+
+### OpenClaw
+
+Add to your `~/.openclaw/openclaw.json` (requires [MCP adapter](https://github.com/androidStern/openclaw-mcp-adapter)):
+
+```json
+{
+  "plugins": {
+    "openclaw-mcp-adapter": {
+      "servers": {
+        "libreart": {
+          "transport": "stdio",
+          "command": "npx",
+          "args": ["-y", "libreart"],
+          "env": {
+            "MEIGEN_API_TOKEN": "meigen_sk_..."
+          }
+        }
+      }
+    }
+  }
+}
+```
+
+### Any MCP-Compatible Host
+
+Add to your `.mcp.json` or equivalent config:
+
+```json
+{
+  "mcpServers": {
+    "libreart": {
+      "type": "stdio",
+      "command": "npx",
+      "args": ["-y", "libreart"],
+      "env": {
+        "MEIGEN_API_TOKEN": "meigen_sk_..."
+      }
+    }
+  }
+}
+```
+
+> **Tip:** Even without an API key, free features (inspiration search, prompt enhancement, model listing) work immediately.
+
+---
+
+## What Can It Do
+
+### 1. Explore & Get Inspired
+
+> "Help me think of something for a social media post"
+
+When you're not sure what to create, the assistant browses the curated gallery of 1,300+ trending prompts, shows you visual previews, and lets you pick a style before generating anything.
+
+### 2. Simple Idea to Pro Image
+
+> "A cat sitting in a Japanese garden"
+
+Short descriptions get automatically enhanced into professional prompts with specific visual details — lighting, composition, lens, color palette — then generated. No prompt engineering knowledge needed.
+
+### 3. Batch & Variant Generation
+
+> "Design 4 different logo concepts for a coffee brand"
+
+The assistant writes distinct creative directions for each variant, generates them in parallel (up to 4 at once), and presents the results with comparative commentary.
+
+### 4. Multi-Step Creative Workflow
+
+> "Create a logo, then show me how it looks on a mug and a t-shirt"
+
+The assistant chains generation steps: creates the base image first, then uses it as a reference to generate product mockups — maintaining visual consistency across the series.
+
+### 5. Reference Image Style Transfer
+
+> "Use this local photo as a style reference for my new image"
+
+Local images are automatically compressed and uploaded. The returned URL works as a reference across all three providers — for style transfer, composition guidance, or img2img workflows.
+
+---
+
+<h2 id="tools">Tools</h2>
+
+| Tool | Free | Description |
+|------|------|-------------|
+| `search_gallery` | Yes | Search 1,300+ curated trending prompts with visual previews |
+| `get_inspiration` | Yes | Get full prompt, all images, and metadata for any gallery entry |
+| `enhance_prompt` | Yes | Transform a brief idea into a professional image prompt (runs locally) |
+| `list_models` | Yes | List available models across all configured providers |
+| `upload_reference_image` | Yes | Compress and upload a local image for use as a reference |
+| `comfyui_workflow` | Yes | Manage ComfyUI workflow templates: list, view, import, modify, delete |
+| `generate_image` | Key | Generate an image — routes to the best available provider automatically |
+
+---
+
+<h2 id="providers">Providers</h2>
+
+LibreArt supports three image generation backends. You can configure one or multiple — the assistant auto-selects the best available, or you can specify per-request.
+
+### MeiGen Platform (Recommended)
+
+Cloud API with multiple model options: NanoBanana Pro, Seedream 4.5, Midjourney Niji7, and more. Free daily credits included.
+
+```json
+{ "meigenApiToken": "meigen_sk_..." }
+```
+
+### ComfyUI (Local)
+
+Run generation on your own GPU with full control over models, samplers, and workflow parameters. Import any ComfyUI API-format workflow.
+
+```json
+{
+  "comfyuiUrl": "http://localhost:8188",
+  "comfyuiDefaultWorkflow": "txt2img"
+}
+```
+
+### OpenAI-Compatible API
+
+Bring your own API key for OpenAI (gpt-image-1), Together AI, Fireworks AI, or any OpenAI-compatible service.
+
+```json
+{
+  "openaiApiKey": "sk-...",
+  "openaiBaseUrl": "https://api.openai.com",
+  "openaiModel": "gpt-image-1"
+}
+```
+
+> All three providers support **reference images**. MeiGen and OpenAI accept URLs directly; ComfyUI injects them into LoadImage nodes in your workflow.
+
+---
+
+## Configuration
+
+### Interactive Setup (Recommended)
+
+```
+/meigen:setup
+```
+
+The wizard walks you through provider selection, API key entry, and ComfyUI workflow import. You can also paste a `curl` command from your API provider's docs — it auto-extracts the key, URL, and model.
+
+### Config File
+
+Configuration is stored at `~/.config/meigen/config.json`. ComfyUI workflows are stored at `~/.config/meigen/workflows/`.
+
+### Environment Variables
+
+Environment variables take priority over the config file.
+
+| Variable | Description |
+|----------|-------------|
+| `MEIGEN_API_TOKEN` | MeiGen platform token |
+| `OPENAI_API_KEY` | OpenAI / compatible API key |
+| `OPENAI_BASE_URL` | API base URL (default: `https://api.openai.com`) |
+| `OPENAI_MODEL` | Default model (default: `gpt-image-1`) |
+| `COMFYUI_URL` | ComfyUI server URL (default: `http://localhost:8188`) |
+
+---
+
+## Development
+
+```bash
+npm install        # Install dependencies
+npm run build      # Compile TypeScript → dist/
+npm run dev        # Development mode (tsx)
+npm run typecheck  # Type check without emitting
+```
+
+---
+
+## License
+
+MIT

package/README.zh-CN.md

@@ -0,0 +1,242 @@
+<p align="center">
+  <img src="assets/banner.jpg" alt="LibreArt Banner" width="600">
+</p>
+
+<h1 align="center">
+  LibreArt — AI 视觉创意专家
+</h1>
+
+<p align="center">
+  <strong>让你的 AI 助手成为图片生成的创意总监</strong><br>
+  <sub>适用于 Claude Code、OpenClaw 及任何 MCP 兼容客户端的插件</sub>
+</p>
+
+<p align="center">
+  <a href="https://www.meigen.ai"><img src="https://img.shields.io/badge/Gallery-meigen.ai-blue?style=flat-square" alt="Gallery"></a>
+  <a href="#工具"><img src="https://img.shields.io/badge/Tools-7-green?style=flat-square" alt="7 Tools"></a>
+  <a href="#生成后端"><img src="https://img.shields.io/badge/Providers-3-orange?style=flat-square" alt="3 Providers"></a>
+  <a href="LICENSE"><img src="https://img.shields.io/badge/License-MIT-lightgrey?style=flat-square" alt="MIT"></a>
+</p>
+
+<p align="center">
+  <a href="#安装">安装</a> •
+  <a href="#它能做什么">它能做什么</a> •
+  <a href="#生成后端">生成后端</a> •
+  <a href="#配置">配置</a>
+</p>
+
+<p align="center">
+  <a href="README.md">English</a> | <strong>中文</strong>
+</p>
+
+---
+
+## 为什么选择 LibreArt？
+
+大多数 AI 图片工具，要么是简单的"输入提示词 → 输出图片"的 API，要么是需要专业知识的复杂界面。
+
+LibreArt 走了一条不同的路：它赋予你的 AI 助手**专业的创意能力**——1,300+ 精选热门提示词、风格化的提示词增强技术、多步骤工作流编排——你只需要用自然语言描述想要什么，就能得到专业品质的结果。
+
+**不需要任何提示词工程知识。** 像跟创意总监对话一样，跟你的 AI 助手聊就行了。
+
+---
+
+## 安装
+
+### Claude Code
+
+一行命令，无需 clone 或构建：
+
+```bash
+claude mcp add --transport stdio libreart -- npx -y meigen
+```
+
+安装时传入 API Key：
+
+```bash
+claude mcp add --transport stdio libreart \
+  --env MEIGEN_API_TOKEN=meigen_sk_xxx \
+  -- npx -y meigen
+```
+
+也可以安装后通过交互式向导配置：
+
+```
+/meigen:setup
+```
+
+### OpenClaw
+
+添加到 `~/.openclaw/openclaw.json`（需要 [MCP adapter](https://github.com/androidStern/openclaw-mcp-adapter)）：
+
+```json
+{
+  "plugins": {
+    "openclaw-mcp-adapter": {
+      "servers": {
+        "libreart": {
+          "transport": "stdio",
+          "command": "npx",
+          "args": ["-y", "libreart"],
+          "env": {
+            "MEIGEN_API_TOKEN": "meigen_sk_..."
+          }
+        }
+      }
+    }
+  }
+}
+```
+
+### 其他 MCP 兼容客户端
+
+添加到 `.mcp.json` 或对应配置文件：
+
+```json
+{
+  "mcpServers": {
+    "libreart": {
+      "type": "stdio",
+      "command": "npx",
+      "args": ["-y", "libreart"],
+      "env": {
+        "MEIGEN_API_TOKEN": "meigen_sk_..."
+      }
+    }
+  }
+}
+```
+
+> **提示：** 即使没有 API Key，免费功能（灵感搜索、提示词增强、模型列表）也可以直接使用。
+
+---
+
+## 它能做什么
+
+### 1. 探索灵感
+
+> "帮我想想社交媒体发什么图"
+
+当你还没有明确想法时，助手会浏览 1,300+ 精选热门提示词库，展示视觉预览，让你先挑选喜欢的风格，再开始生成。
+
+### 2. 简单想法变专业图片
+
+> "一只猫坐在日式庭院里"
+
+简短的描述会被自动增强为专业提示词——补充光影、构图、镜头、配色等视觉细节——然后直接生成。不需要任何提示词工程知识。
+
+### 3. 批量变体生成
+
+> "设计 4 个不同风格的咖啡品牌 logo"
+
+助手会为每个变体撰写不同的创意方向，最多 4 张并行生成，并附上对比点评。
+
+### 4. 多步骤创意工作流
+
+> "先做一个 logo，然后看看印在马克杯和 T 恤上的效果"
+
+助手会串联生成步骤：先生成基础图片，再将其作为参考图生成产品效果图——在整个系列中保持视觉一致性。
+
+### 5. 参考图风格迁移
+
+> "用这张本地照片的风格来生成新图片"
+
+本地图片会自动压缩上传，返回的 URL 可用于所有三种后端——风格迁移、构图参考或 img2img 工作流。
+
+---
+
+<h2 id="工具">工具</h2>
+
+| 工具 | 免费 | 说明 |
+|------|------|------|
+| `search_gallery` | 是 | 搜索 1,300+ 精选热门提示词，附带视觉预览 |
+| `get_inspiration` | 是 | 获取某条提示词的完整内容、所有图片和元数据 |
+| `enhance_prompt` | 是 | 将简短想法转化为专业图片提示词（本地运行） |
+| `list_models` | 是 | 列出所有已配置后端的可用模型 |
+| `upload_reference_image` | 是 | 压缩并上传本地图片，用作参考图 |
+| `comfyui_workflow` | 是 | 管理 ComfyUI 工作流模板：列表、查看、导入、修改、删除 |
+| `generate_image` | 需要 Key | 生成图片——自动路由到最佳可用后端 |
+
+---
+
+<h2 id="生成后端">生成后端</h2>
+
+LibreArt 支持三种图片生成后端，可以配置一个或多个——助手会自动选择最佳可用后端，也可以在每次请求时手动指定。
+
+### MeiGen 平台（推荐）
+
+云端 API，支持多种模型：NanoBanana Pro、Seedream 4.5、Midjourney Niji7 等。包含每日免费额度。
+
+```json
+{ "meigenApiToken": "meigen_sk_..." }
+```
+
+### ComfyUI（本地）
+
+在自己的 GPU 上运行，完全控制模型、采样器和工作流参数。支持导入任意 ComfyUI API 格式的工作流。
+
+```json
+{
+  "comfyuiUrl": "http://localhost:8188",
+  "comfyuiDefaultWorkflow": "txt2img"
+}
+```
+
+### OpenAI 兼容 API
+
+使用你自己的 API Key 接入 OpenAI（gpt-image-1）、Together AI、Fireworks AI 或任何 OpenAI 兼容服务。
+
+```json
+{
+  "openaiApiKey": "sk-...",
+  "openaiBaseUrl": "https://api.openai.com",
+  "openaiModel": "gpt-image-1"
+}
+```
+
+> 三种后端都支持**参考图**。MeiGen 和 OpenAI 直接接受 URL；ComfyUI 会将参考图注入到工作流的 LoadImage 节点中。
+
+---
+
+## 配置
+
+### 交互式配置（推荐）
+
+```
+/meigen:setup
+```
+
+配置向导会引导你选择后端、输入 API Key、导入 ComfyUI 工作流。你也可以直接粘贴 API 提供商文档中的 `curl` 命令——自动提取 Key、URL 和模型名。
+
+### 配置文件
+
+配置存储在 `~/.config/meigen/config.json`。ComfyUI 工作流存储在 `~/.config/meigen/workflows/`。
+
+### 环境变量
+
+环境变量优先级高于配置文件。
+
+| 变量 | 说明 |
+|------|------|
+| `MEIGEN_API_TOKEN` | MeiGen 平台 Token |
+| `OPENAI_API_KEY` | OpenAI / 兼容 API Key |
+| `OPENAI_BASE_URL` | API 地址（默认：`https://api.openai.com`） |
+| `OPENAI_MODEL` | 默认模型（默认：`gpt-image-1`） |
+| `COMFYUI_URL` | ComfyUI 服务地址（默认：`http://localhost:8188`） |
+
+---
+
+## 开发
+
+```bash
+npm install        # 安装依赖
+npm run build      # 编译 TypeScript → dist/
+npm run dev        # 开发模式（tsx）
+npm run typecheck  # 类型检查
+```
+
+---
+
+## 许可证
+
+MIT

package/bin/meigen-mcp.js

@@ -0,0 +1,2 @@
+#!/usr/bin/env node
+require('../dist/index.js');