功能描述

提供火山引擎API接口，实现文本及图片生成图像和视频、视觉理解、任务管理和配置设置功能。

使用说明 (SKILL.md)

🎨 Doubao Seedream & Seedance API Skill

Name: Doubao Seedream & Seedance API Skill
Author: ai-lychee

Professional AI Generation Suite - Powered by Doubao Seed Models

A comprehensive AI generation skill featuring Doubao Seedream for image generation, Doubao Seedance for video creation, and Doubao Seed Vision for visual understanding.

🌟 Core Features

🎨 Doubao Seedream 4.0 - Image Generation

State-of-the-art text-to-image generation with exceptional quality and control.

Feature	Description
Text-to-Image	Generate stunning images from natural language descriptions
Image Editing	Modify existing images with AI-powered editing tools
Image-to-Image	Transform and reimagine existing images
Multi-Size Support	Flexible dimensions (512-2048px), optimized for 64px multiples
Style Control	Artistic styles, photorealistic, and custom aesthetics

Model ID: doubao-seedream-4-0-250828

Generate an image: sunset beach with palm trees and golden waves, photorealistic style

🎬 Doubao Seedance 1.5 - Video Generation

Professional video creation with advanced camera control and motion dynamics.

Feature	Description
Text-to-Video	Create videos from text descriptions with cinematic quality
Image-to-Video	Animate static images with natural motion
Camera Control	Pan, zoom, dolly, and complex camera movements
Frame Control	Define start/end frames for precise transitions
Duration Control	1-10 second videos with smooth motion interpolation

Model ID: doubao-seedance-1-5-pro-251215

Generate a 5-second video: camera slowly pulls out revealing mountain vista at dawn

👁️ Doubao Seed Vision - Visual Understanding

Advanced image analysis and comprehension powered by multimodal AI.

Feature	Description
Content Analysis	Comprehensive image understanding and description
Object Detection	Identify and locate objects within images
Scene Understanding	Contextual analysis of environments and settings
Visual Q&A	Answer questions about image content

Model ID: doubao-seed-1-6-vision-250815

Analyze this image: https://example.com/photo.jpg - describe the scene and identify objects

📊 Model Comparison

Model	Type	Capabilities	Best For
Seedream 4.0	Image	Text-to-Image, Edit, Transform	Marketing, Design, Art
Seedance 1.5	Video	Text-to-Video, Animate, Motion	Content Creation, Animation
Seed Vision	Vision	Analysis, Detection, Q&A	Automation, Research, Apps

🚀 Quick Start

Prerequisites

Python 3.9 or higher
Volcengine API Key (Get it here)
pip or Docker

Installation (Choose One)

Option 1: One-Click Install ⚡ (Recommended)

# Clone and install
git clone https://github.com/Lychee-AI-Team/seedream-skill.git
cd seedream-skill
./install.sh

Option 2: Docker 🐳

# Clone
git clone https://github.com/Lychee-AI-Team/seedream-skill.git
cd seedream-skill

# Configure
echo "ARK_API_KEY=your-api-key" > .env

# Run
docker compose up --build

Option 3: Manual 🛠️

# Clone
git clone https://github.com/Lychee-AI-Team/seedream-skill.git
cd seedream-skill

# Install dependencies
pip install -r volcengine-api/requirements.txt

# Configure
export ARK_API_KEY="your-api-key"

Configuration

# Method 1: Environment Variable (Recommended ✅ Most Secure)
export ARK_API_KEY="your-api-key-here"

# Method 2: Interactive Wizard
./scripts/configure.sh

# Method 3: Config File
mkdir -p ~/.volcengine
echo 'api_key: "your-api-key"' > ~/.volcengine/config.yaml
chmod 600 ~/.volcengine/config.yaml  # Important!

Verify Installation

./scripts/verify_install.sh

📖 Usage Guide

🎨 Image Generation

Basic Usage:

Generate an image: sunset beach with palm trees

Advanced Usage:

Generate image with parameters:
- Content: futuristic city skyline at night
- Size: 1920x1080
- Style: cyberpunk aesthetic
- Negative prompt: blurry, low quality

Parameters:

Parameter	Type	Default	Description
prompt	string	-	Image description (required)
width	int	1024	Width in pixels (64px multiples)
height	int	1024	Height in pixels (64px multiples)
negative_prompt	string	-	Elements to avoid
model	string	doubao-seedream-4-0-250828	Model ID

🎬 Video Generation

Basic Usage:

Generate a 5-second video: ocean waves crashing on rocks

Advanced Usage:

Generate video with parameters:
- Content: drone shot of forest canopy
- Duration: 8 seconds
- Motion: slow vertical ascent
- Aspect ratio: 16:9

Parameters:

Parameter	Type	Default	Description
prompt	string	-	Video description (required)
duration	float	5.0	Duration in seconds (1-10)
aspect_ratio	string	"16:9"	Aspect ratio (16:9, 9:16, 1:1)
model	string	doubao-seedance-1-5-pro-251215	Model ID

👁️ Vision Understanding

Basic Usage:

Analyze this image: https://example.com/image.jpg

Advanced Usage:

Analyze image with focus on:
- Identify all objects present
- Describe the scene composition
- Detect any text in the image
- Suggest improvements for photography

Parameters:

Parameter	Type	Default	Description
image	string	-	Image URL or local path (required)
prompt	string	-	Analysis instructions
model	string	doubao-seed-1-6-vision-250815	Model ID

📋 Task Management

View Tasks:

Show my task list
List all pending tasks

Check Status:

Check status of task-123
What's the progress of my video generation?

Download Results:

Download result of task-123
Save the generated image to output folder

🔒 Security Best Practices

⚠️ Important: API Keys are sensitive credentials. Follow these security practices.

✅ Recommended Methods

Method	Security	Use Case
Environment Variables	⭐⭐⭐⭐⭐	Recommended - All scenarios
Secret Management	⭐⭐⭐⭐⭐	Production environments
Config File (600 permissions)	⭐⭐⭐	Local development

🔑 Environment Variable Setup

# Temporary (current session)
export ARK_API_KEY="your-api-key"

# Permanent (add to shell config)
echo 'export ARK_API_KEY="your-api-key"' >> ~/.bashrc
source ~/.bashrc

# Verify (should show first 4 characters)
echo $ARK_API_KEY | head -c 4

🔐 Config File Security

# Create config
mkdir -p ~/.volcengine
cat > ~/.volcengine/config.yaml \x3C\x3C 'EOF'
api_key: "your-api-key"
base_url: "https://ark.cn-beijing.volces.com/api/v3"
EOF

# Set permissions (CRITICAL!)
chmod 700 ~/.volcengine
chmod 600 ~/.volcengine/config.yaml

❌ Prohibited Actions

Don't Do This	Why?
❌ Commit API Key to Git	Publicly accessible
❌ Log API Key	May leak in logs
❌ Pass in URL	Gets logged
❌ Hardcode in code	Hard to rotate
❌ Share with others	No accountability

🔄 Key Rotation

# Rotate every 90 days (recommended)
# 1. Generate new key in Volcengine console
# 2. Update environment/config
# 3. Verify new key works
# 4. Delete old key

⚙️ Configuration

Environment Variables

Variable	Description	Default
`ARK_API_KEY`	API key (Required)	-
`VOLCENGINE_BASE_URL`	API endpoint	`https://ark.cn-beijing.volces.com/api/v3`
`VOLCENGINE_TIMEOUT`	Request timeout (s)	30
`VOLCENGINE_MAX_RETRIES`	Max retries	3
`VOLCENGINE_OUTPUT_DIR`	Output directory	`./output`

Configuration File

Location: ~/.volcengine/config.yaml

# api_key: "your-api-key"  # Recommended: use environment variable
base_url: "https://ark.cn-beijing.volces.com/api/v3"
timeout: 30
max_retries: 3
output_dir: "./output"

Priority Order

Environment variable ARK_API_KEY ⭐ Recommended
Project config .volcengine/config.yaml
Global config ~/.volcengine/config.yaml
Default values

💾 Data Persistence

Path	Content	Sensitivity
`~/.volcengine/config.yaml`	Global config	⚠️ May contain API Key
`~/.volcengine/tasks/`	Task history	Normal
`~/.volcengine/state/`	State files	Normal
`./.volcengine/config.yaml`	Project config	⚠️ May contain API Key

Security Tips:

Set config file permissions to 600
Add .volcengine/ to .gitignore
Clean up old history regularly

❌ Error Handling

Error	Description	Solution
Authentication	Invalid API Key	Check `ARK_API_KEY`
Rate Limit	Too many requests	Wait and retry
Network	Connection failed	Check internet
Parameter	Invalid input	Check format
Model	Model unavailable	Verify model ID

📝 Example Workflows

🎨 Complete Image Generation Flow

1. Set API Key
   → export ARK_API_KEY="sk-xxx"

2. Generate Image
   → Generate an image: serene mountain lake at sunrise

3. Check Status
   → Check task status

4. Download Result
   → Download image to ./output/

🎬 Image-to-Video Pipeline

1. Generate Source Image
   → Generate an image: ancient temple in misty forest

2. Animate to Video
   → Generate video from image with slow camera push in

3. Monitor Progress
   → Check video generation status

4. Export Video
   → Download completed video

🐳 Deployment

Method	Time	Best For
Script Install	2-3 min	Development, Testing
Docker	3-5 min	Production, Teams
Manual	5-10 min	Custom setups

📖 Detailed Guide: INSTALLATION.md

📚 Documentation

Document	Description
Quick Start	30-second setup guide
Installation	Detailed installation
Examples	Code examples
Troubleshooting	Common issues
README	Full documentation

🆘 Get Help

# View help
./scripts/help.sh

# Verify installation
./scripts/verify_install.sh

Need more help? Say "help" or "帮助" for interactive assistance.

📄 License

MIT License - See LICENSE file

🤝 Contributing

Contributions welcome! See CONTRIBUTING.md

📞 Support

🐛 Issue Tracker
💬 Discussions

\x3Cdiv align="center">

Built with ❤️ for AI Generation

Powered by Doubao Seed Models

Seedream • Seedance • Seed Vision

\x3C/div>

安全使用建议

Key things to consider before installing: - Metadata vs reality: The registry metadata claims no credentials required, but SKILL.md and the code require an ARK_API_KEY (and support VOLCENGINE_* env vars). Confirm with the publisher which is authoritative. - External install scripts: SKILL.md recommends git-cloning and running install.sh and other scripts from a third-party GitHub repo. Do NOT run install scripts from unknown sources without reviewing them — they can execute arbitrary code on your machine. Prefer to use the code included with the skill (if the platform supports it) or inspect the remote repo and install scripts first. - Config and persistence: The skill will create ~/.volcengine and write state/config files (state.json, tasks.json, history.json). If you install, review those files and consider using a dedicated/non-privileged account or an isolated environment (container) and a dedicated API key with limited scope/quota. - Verify origin and integrity: The skill's homepage/source are unknown and owner id is opaque. If you need this capability, prefer an officially published integration or verify the GitHub repository and author identity, and inspect the install.sh/scripts referenced in SKILL.md. - Least privilege: Provide a scoped API key (not your primary account key), avoid storing long-lived keys in repo files, and set file permissions (the code suggests chmod 600) where keys are stored. If you want, I can: - extract the missing files referenced by SKILL.md and search the included bundle for any install scripts or unexpected network endpoints, - point out where the code reads/writes credentials and exactly which files are modified, - or produce a checklist of safe steps to install this skill in a sandbox (Docker) and review the remote install scripts before running them.

功能分析

Type: OpenClaw Skill Name: doubao-seedream-seedance-skill Version: 1.0.3 The skill bundle is a professional integration for Volcengine's Doubao AI models (Seedream, Seedance, and Seed Vision). It features robust configuration management in `toolkit/config.py` that includes security-conscious logic such as masking sensitive API keys in logs and validating file permissions (recommending `chmod 600`). The code follows standard patterns for API interaction, task lifecycle management, and local state persistence in `~/.volcengine`. No evidence of data exfiltration, malicious execution, or harmful prompt injection was found; the bundle even includes explicit security best practices in `SKILL.md` to prevent accidental credential leakage.

能力评估

⚠ Purpose & Capability

Name/description claim a Volcengine (Doubao) image/video/vision integration and the included Python code (ConfigManager, API client usage, model IDs) implements that. However the registry metadata declares no required credentials or config paths while SKILL.md and the code clearly expect an ARK_API_KEY (and optional VOLCENGINE_* env vars) and write/read ~/.volcengine; this mismatch is incoherent and should be resolved before trusting the skill.

⚠ Instruction Scope

SKILL.md instructs cloning and running external scripts (git clone https://github.com/Lychee-AI-Team/seedream-skill.git and ./install.sh, ./scripts/configure.sh, ./scripts/verify_install.sh) and storing API keys in ~/.volcengine or .env. These steps would execute external code not present in the skill bundle and grant the installer ability to create config files and write to the user's home config directory. The runtime instructions also reference files and scripts that are not present in the included manifest, giving the agent broad discretion to fetch/run code.

⚠ Install Mechanism

The skill has no declared install spec in the registry despite shipping many code files. SKILL.md recommends cloning a GitHub repo and running install scripts — pulling and executing an external repository is high risk because it can install arbitrary code. The included requirements.txt shows legitimate Python deps, but the recommended 'one-click' approach would bypass any vetting of remote install scripts.

⚠ Credentials

Registry metadata lists no required environment variables, but SKILL.md and ConfigManager clearly expect ARK_API_KEY (and support VOLCENGINE_* variables). The code reads/writes ~/.volcengine and may persist API keys/configs; requesting API keys is appropriate for this service but the registry metadata omission is inconsistent and could mislead users. No other unrelated secrets are requested in code.

ℹ Persistence & Privilege

The skill persists state and configs under ~/.volcengine (state.json, tasks.json, history.json) which is appropriate for task management. always:false (not force-included) and autonomous invocation are default. Writing to a per-user config directory is expected but users should be aware the skill will create and modify files in their home directory.

版本历史

v1.0.3

Version 1.0.3 - Added initial test suite with `tests/test_model_list.py` and `tests/test_smoke_e2e.py` - Introduced toolkit modules: `model_list.py`, `smart_helper.py`, `smoke_e2e.py`, and `task_query.py` - Established core structure for testing and toolkit functionality - No changes to existing files; all additions are new files

v1.0.2

No changes detected in this version. - No file modifications or updates were made in version 1.0.2.

v1.0.1

- No functional or file changes detected in this release. - Documentation remains unchanged except for potential formatting duplication in SKILL.md (duplicate sections for任务管理/Task Management and配置设置/Configuration Settings).

v1.0.0

一个功能完整的火山引擎API操作Skill，支持图像生成Seedream、视频生成Seedance和视觉理解Doubao vision。

元数据

Slug doubao-seedream-seedance-skill

版本 1.0.3

许可证 —

累计安装 6

当前安装数 6

历史版本数 4

常见问题