@optima-chat/comfy-cli 0.5.0 → 0.6.0

package/.claude/settings.local.json

@@ -31,7 +31,10 @@
       "Bash(gh issue view:*)",
       "Bash(python3:*)",
       "Bash(git reset:*)",
-      "Bash(git checkout:*)"
+      "Bash(git checkout:*)",
+      "Bash(git branch:*)",
+      "Bash(gh run view:*)",
+      "WebFetch(domain:code.claude.com)"
     ],
     "deny": [],
     "ask": []

package/.claude/skills/comfy-cli/SKILL.md

@@ -3,439 +3,144 @@ name: comfy-cli
 description: "ComfyUI CLI tool for AI agents. ALWAYS use when user wants to: generate images (生成图片/画图/图像/生成画), edit images (编辑图片/修改图片/图生图/改图), create videos (生成视频/图生视频/制作视频), manage ComfyUI workflows (工作流/ComfyUI). Uses 'comfy generate', 'comfy edit', 'comfy video' commands."
 ---
 
-# ComfyUI CLI - AI Image & Video Generation
+# ComfyUI CLI
 
-## When to Use This Skill
+## Instructions
 
-Use this skill when:
-- User wants to generate images from text prompts (文生图)
-- User wants to edit or transform existing images (图生图)
-- User wants to create videos from images (图生视频)
-- User asks about ComfyUI workflows, queue status, or models
-- User mentions: 生成图片、画图、图像处理、视频生成、ComfyUI
+### 核心命令
 
-## Available Commands
-
-### 1. Generate Images: `comfy generate`
-
-Generate images from text prompts using Stable Diffusion / FLUX models.
-
-**Syntax:**
+**生成图像：**
 ```bash
-comfy generate "PROMPT" [OPTIONS]
+comfy generate "提示词" [--width 1024] [--height 1024] [--no-wait] [--pretty]
 ```
+- 从文本提示词生成图像
+- 默认等待完成并显示实时进度，返回 JSON
+- `--no-wait`: 提交后立即返回
+- `--pretty`: 人类可读的表格格式
 
-**Parameters:**
-- `PROMPT`: Text description of the image to generate
-- `--width <number>`: Image width (default: 1024)
-- `--height <number>`: Image height (default: 1024)
-- `--no-wait`: Submit and return immediately without waiting for completion
-- `--pretty`: Human-readable table format output (default: JSON)
-
-**Returns:**
-- JSON with `prompt_id`, `status`, `outputs` (image file paths)
-- Default behavior: waits for completion and shows real-time progress
-
-**When to use:**
-- User asks to generate/create images from text
-- User provides image description
-- Text-to-image generation
-
-### 2. Edit Images: `comfy edit`
-
-Edit or transform existing images with text prompts (Image-to-Image).
-
-**Syntax:**
+**编辑图像：**
 ```bash
-comfy edit <IMAGE_PATH> "PROMPT" [OPTIONS]
+comfy edit <图像路径> "提示词" [--no-wait] [--pretty]
 ```
+- 使用文本提示词编辑或转换现有图像
+- 支持风格转换、细节增强等
 
-**Parameters:**
-- `IMAGE_PATH`: Path to input image file
-- `PROMPT`: Transformation description (e.g., "anime style", "add more details")
-- `--no-wait`: Submit and return immediately without waiting for completion
-- `--pretty`: Human-readable table format output (default: JSON)
-
-**Returns:**
-- JSON with `prompt_id`, `status`, `outputs` (edited image file paths)
-- Default behavior: waits for completion and shows real-time progress
-
-**When to use:**
-- User wants to modify/edit an existing image
-- User wants to apply style transfer
-- Image-to-image transformation
-
-### 3. Create Videos: `comfy video`
-
-Generate videos from images using I2V (Image-to-Video) models.
-
-**Syntax:**
+**生成视频：**
 ```bash
-comfy video <IMAGE_PATH> [OPTIONS]
+comfy video <图像路径> [-p "运动描述"] [-n "负面提示"] [--width 512] [--height 512] [--no-wait] [--pretty]
 ```
+- 从图像生成视频（图生视频）
+- 可选运动描述和负面提示
 
-**Parameters:**
-- `IMAGE_PATH`: Path to input image file
-- `-p, --prompt <text>`: Motion description prompt (optional)
-- `-n, --negative <text>`: Negative prompt (optional)
-- `--width <number>`: Video width (default: 512)
-- `--height <number>`: Video height (default: 512)
-- `--no-wait`: Submit and return immediately without waiting for completion
-- `--pretty`: Human-readable table format output (default: JSON)
-
-**Returns:**
-- JSON with `prompt_id`, `status`, `outputs` (video file paths)
-- Default behavior: waits for completion and shows real-time progress
-
-**When to use:**
-- User wants to animate a static image
-- User wants to create video from image
-- Image-to-video generation
-
-### 4. Download Results: `comfy download`
-
-Download generated images/videos to local directory.
-
-**Syntax:**
+**下载结果：**
 ```bash
-comfy download <PROMPT_ID> [OPTIONS]
+comfy download <PROMPT_ID> [-o ./comfy-output] [--pretty]
 ```
+- 下载生成的图像/视频到本地目录
 
-**Parameters:**
-- `PROMPT_ID`: The prompt ID returned from generate/edit/video commands
-- `-o, --output <dir>`: Output directory (default: ./comfy-output)
-- `--pretty`: Human-readable output
+### 工作流管理
 
-**Returns:**
-- JSON with downloaded file paths
-
-### 5. Workflow Management
-
-**List workflow history:**
 ```bash
-comfy workflow list [--limit 10] [--pretty]
+comfy workflow list [--limit 10] [--pretty]          # 列出历史
+comfy workflow get <PROMPT_ID> [--pretty]            # 获取结果
+comfy workflow submit <工作流文件> [--pretty]         # 提交自定义工作流
 ```
 
-**Get workflow result:**
-```bash
-comfy workflow get <PROMPT_ID> [--pretty]
-```
+### 队列管理
 
-**Submit custom workflow:**
 ```bash
-comfy workflow submit <WORKFLOW_FILE> [--pretty]
+comfy queue status [--pretty]                        # 查看队列状态
+comfy queue clear --confirm [--pretty]               # 清空队列
+comfy queue delete <项目ID> [--pretty]               # 删除队列项
+comfy interrupt [--pretty]                           # 中断当前执行
 ```
 
-### 6. Queue Management
+### 模型管理
 
-**Check queue status:**
 ```bash
-comfy queue status [--pretty]
+comfy model list [--type checkpoints|loras|vae|controlnet|upscale_models] [--pretty]
 ```
 
-**Clear queue:**
-```bash
-comfy queue clear --confirm [--pretty]
-```
-
-**Delete queue item:**
-```bash
-comfy queue delete <ITEM_ID> [--pretty]
-```
-
-**Interrupt current execution:**
-```bash
-comfy interrupt [--pretty]
-```
-
-### 7. Model Management
-
-**List available models:**
-```bash
-comfy model list [--type checkpoints|loras|vae|...] [--pretty]
-```
+### 系统信息
 
-**Model types:**
-- `checkpoints`: Stable Diffusion / FLUX checkpoint models
-- `loras`: LoRA models
-- `vae`: VAE models
-- `controlnet`: ControlNet models
-- `upscale_models`: Upscaler models
-
-### 8. System Information
-
-**Get system stats:**
 ```bash
-comfy system stats [--pretty]
+comfy system stats [--pretty]                        # GPU/CPU/内存状态
 ```
 
-Returns: GPU/CPU usage, memory, device info
-
-### 9. Configuration
-
-**Set ComfyUI server address (required for first use):**
-```bash
-comfy config set server http://localhost:8188
-```
+### 配置
 
-**View current config:**
 ```bash
-comfy config list [--pretty]
-```
-
-**Other config options:**
-```bash
-comfy config set timeout 30000
-comfy config set outputDir ./my-outputs
-```
-
-## Output Format
-
-All commands default to **JSON output** (AI-friendly, reduces 50%+ tokens):
-
-```json
-{
-  "success": true,
-  "data": {
-    "prompt_id": "abc123",
-    "status": "completed",
-    "outputs": ["./comfy-output/image_001.png"]
-  }
-}
+comfy config set server http://localhost:8188        # 设置服务器地址（首次必需）
+comfy config list [--pretty]                         # 查看当前配置
 ```
 
-Use `--pretty` flag for human-readable table format with colors.
+### 输出格式
 
-## Workflow Examples
+- **默认**：JSON 格式（AI 友好，减少 token）
+- **--pretty**：带颜色的表格格式（人类可读）
 
-### Example 1: Quick Image Generation
-**User**: "帮我生成一张猫的图片"
+## Examples
 
-**Assistant**:
+### 示例 1：生成图像
 ```bash
 comfy generate "a cute cat"
 ```
+返回包含 `prompt_id` 和 `outputs`（图像文件路径）的 JSON
 
-**Result**: Returns JSON with `prompt_id` and `outputs` containing image file paths. Default behavior waits for completion and shows real-time progress via WebSocket.
-
-### Example 2: High-Resolution Image
-**User**: "生成一张 1920x1080 的风景画"
-
-**Assistant**:
+### 示例 2：高分辨率图像
 ```bash
-comfy generate "beautiful landscape with mountains and lake" --width 1920 --height 1080
+comfy generate "beautiful landscape with mountains" --width 1920 --height 1080
 ```
 
-### Example 3: Image Style Transfer
-**User**: "把这张图片改成动漫风格"
-
-**Assistant** (assuming user uploaded `photo.jpg`):
+### 示例 3：图像风格转换
 ```bash
 comfy edit photo.jpg "anime style"
 ```
 
-### Example 4: Image Editing with Details
-**User**: "给这张照片添加更多细节，提高质量"
-
-**Assistant**:
-```bash
-comfy edit photo.jpg "add more details, enhance quality, 4k resolution"
-```
-
-### Example 5: Video Creation
-**User**: "用这张人物图片生成一个视频"
-
-**Assistant**:
+### 示例 4：视频生成
 ```bash
 comfy video portrait.png --prompt "smooth motion, natural movement"
 ```
 
-### Example 6: Non-blocking Generation
-**User**: "生成一张图片，不用等完成"
-
-**Assistant**:
-```bash
-comfy generate "sunset over ocean" --no-wait
-```
-
-Returns immediately with `prompt_id`. Check status later with:
-```bash
-comfy workflow get <prompt_id>
-```
-
-### Example 7: Batch Image Generation
-**User**: "帮我生成 3 张不同的猫的图片"
-
-**Assistant**:
-```bash
-comfy generate "a cute cat sitting" --no-wait
-comfy generate "a cat playing with toys" --no-wait
-comfy generate "a cat sleeping on a couch" --no-wait
-```
-
-Then check queue status:
+### 示例 5：批量生成（非阻塞）
 ```bash
+comfy generate "a cat sitting" --no-wait
+comfy generate "a cat playing" --no-wait
+comfy generate "a cat sleeping" --no-wait
 comfy queue status
 ```
 
-### Example 8: Workflow with Model Check
-**User**: "检查一下有哪些可用的模型，然后生成图片"
-
-**Assistant**:
+### 示例 6：完整工作流
 ```bash
-# First check available models
-comfy model list --type checkpoints
-
-# Then generate with appropriate prompt
-comfy generate "a beautiful portrait"
-```
-
-### Example 9: Complete Workflow
-**User**: "生成一张图片，编辑它，然后生成视频"
-
-**Assistant**:
-```bash
-# 1. Generate base image
+# 1. 生成图像
 comfy generate "a portrait of a woman"
-# Output: ./comfy-output/image_001.png
+# 输出：./comfy-output/image_001.png
 
-# 2. Edit the image
+# 2. 编辑图像
 comfy edit ./comfy-output/image_001.png "add cinematic lighting"
-# Output: ./comfy-output/image_002.png
+# 输出：./comfy-output/image_002.png
 
-# 3. Create video from edited image
-comfy video ./comfy-output/image_002.png --prompt "slow motion, cinematic"
-# Output: ./comfy-output/video_001.mp4
+# 3. 生成视频
+comfy video ./comfy-output/image_002.png --prompt "slow motion"
+# 输出：./comfy-output/video_001.mp4
 ```
 
-## Best Practices
-
-### 1. Output Format Selection
-- **Default (JSON)**: Best for AI agents, programmatic parsing, reduces tokens
-- **Pretty mode**: Best for human inspection, debugging, visual feedback
-- Use `--pretty` when user explicitly wants to see formatted output
-
-### 2. Wait Strategy
-- **Default (wait)**: Recommended for interactive workflows, shows real-time progress
-- **--no-wait**: Use when submitting multiple jobs, batch processing, or time-consuming operations
-- After `--no-wait`, use `comfy workflow get <prompt_id>` to check completion
-
-### 3. Error Handling
-- Always check `success` field in JSON response
-- If error occurs, check:
-  1. ComfyUI server is running: `comfy system stats`
-  2. Server address is correct: `comfy config list`
-  3. Queue is not stuck: `comfy queue status`
-
-### 4. Performance Optimization
-- Use appropriate image sizes (1024x1024 for SD, 512x512 for videos)
-- Consider `--no-wait` for batch operations
-- Check system resources before large generations: `comfy system stats`
-
-### 5. First-time Setup
-Always configure server address before first use:
+### 示例 7：检查状态和下载
 ```bash
-comfy config set server http://localhost:8188
-```
-
-## Common Patterns
+# 非阻塞提交
+comfy generate "sunset" --no-wait
+# 返回：{ "prompt_id": "abc123" }
 
-### Pattern 1: Generate and Download
-```bash
-# Generate image
-comfy generate "a cat" --no-wait
-# Returns: { "prompt_id": "abc123" }
+# 稍后检查状态
+comfy workflow get abc123
 
-# Download result later
+# 下载结果
 comfy download abc123
 ```
 
-### Pattern 2: Iterative Editing
-```bash
-# Start with base image
-comfy generate "a portrait"
-
-# Edit step 1
-comfy edit output_001.png "add dramatic lighting"
-
-# Edit step 2
-comfy edit output_002.png "increase contrast"
-```
-
-### Pattern 3: Queue Management
-```bash
-# Submit multiple jobs
-comfy generate "cat" --no-wait
-comfy generate "dog" --no-wait
-comfy generate "bird" --no-wait
-
-# Check queue
-comfy queue status
-
-# If needed, clear queue
-comfy queue clear --confirm
-```
+### 重要提示
 
-## Tips
-
-**Server Connection:**
-- Default server: `http://localhost:8188`
-- For remote servers: `comfy config set server http://remote-ip:8188`
-- Check connection: `comfy system stats`
-
-**Model Selection:**
-- ComfyUI automatically uses default checkpoint model
-- Check available models: `comfy model list --type checkpoints`
-- For custom workflows, use: `comfy workflow submit custom.json`
-
-**Output Files:**
-- Default output directory: `./comfy-output/`
-- Change with: `comfy config set outputDir ./my-outputs`
-- Downloaded files retain original ComfyUI filenames
-
-**Real-time Progress:**
-- Default wait mode shows progress via WebSocket
-- Progress includes: node execution, current step, percentage
-- Use `--no-wait` to skip progress monitoring
-
-**Interruption:**
-- Stop current generation: `comfy interrupt`
-- Clear pending jobs: `comfy queue clear --confirm`
-
-**JSON Parsing:**
-- All JSON responses follow consistent structure
-- Check `success` field first
-- Access data via `data` field
-- Errors include `error` field with details
-
-## Configuration
-
-Config file location: `~/.comfy-cli/config.json`
-
-**Available settings:**
-- `server`: ComfyUI server address (required)
-- `timeout`: Request timeout in milliseconds (default: 30000)
-- `autoConnect`: Auto-connect to WebSocket (default: true)
-- `outputDir`: Default output directory (default: `./comfy-output`)
-
-## Troubleshooting
-
-**"Connection refused" error:**
-- Ensure ComfyUI is running: Check browser at http://localhost:8188
-- Verify server address: `comfy config list`
-- Test connection: `comfy system stats`
-
-**"Workflow failed" error:**
-- Check error details in JSON response
-- Verify model files are available: `comfy model list`
-- Check ComfyUI console logs
-
-**Slow generation:**
-- Check GPU availability: `comfy system stats`
-- Reduce image resolution
-- Check queue: `comfy queue status`
-
-**Missing output files:**
-- Check output directory: `comfy config list`
-- Verify generation completed: `comfy workflow get <prompt_id>`
-- Use `comfy download <prompt_id>` to re-download
+- 首次使用前必须配置服务器地址：`comfy config set server http://localhost:8188`
+- 默认输出 JSON 格式，适合 AI 解析
+- 使用 `--pretty` 获取人类可读输出
+- 默认等待完成并显示实时进度，使用 `--no-wait` 跳过等待

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@optima-chat/comfy-cli",
-  "version": "0.5.0",
+  "version": "0.6.0",
   "description": "A CLI tool for ComfyUI designed for LLM interactions",
   "type": "module",
   "main": "dist/index.js",