Skill flagged — suspicious patterns detected

ClawHub Security flagged this skill as suspicious. Review the scan results before using.

doubao-seedream-skill

v1.0.2

调用火山引擎 Seedream 图片生成 API。当用户需要生成图片时使用此 skill。

2· 301·0 current·0 all-time

Install

OpenClaw Prompt Flow

Install with OpenClaw

Best for remote or guided setup. Copy the exact prompt, then paste it into OpenClaw for lamuier/doubao-seedream-skill.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "doubao-seedream-skill" (lamuier/doubao-seedream-skill) from ClawHub.
Skill page: https://clawhub.ai/lamuier/doubao-seedream-skill
Keep the work scoped to this skill only.
After install, inspect the skill metadata and help me finish setup.
Use only the metadata you can verify from ClawHub; do not invent missing requirements.
Ask before making any broader environment changes.

Command Line

CLI Commands

Use the direct CLI path if you want to install manually and keep every step visible.

OpenClaw CLI

Canonical install target

openclaw skills install lamuier/doubao-seedream-skill

ClawHub CLI

Package manager switcher

npx clawhub@latest install doubao-seedream-skill
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Suspicious
medium confidence
Purpose & Capability
The skill's name, description, SKILL.md, and code are coherent: they call the Volcengine (Seedream) images API at ark.cn-beijing.volces.com to generate and download images. However the registry metadata listed earlier claims no required environment variables while SKILL.md, .env, and seedream_api.py require VOLCENGINE_API_KEY — a packaging/metadata omission.
Instruction Scope
Runtime instructions and the code stay within the stated purpose: they build API requests, optionally base64-encode provided local image files, send them to the Seedream API, and download returned image URLs into an output directory. Important behavioral notes: the skill will read user-supplied local image files (encode & upload them) and will perform outbound HTTP requests (API calls and downloads of generated image URLs). It does not attempt to read arbitrary other system files or unrelated environment variables.
Install Mechanism
No install spec is provided (instruction-only plus a Python module). This is low-risk from install provenance; dependencies (requests, python-dotenv) are normal Python packages and are referenced in README but not enforced by an install script.
!
Credentials
The code and SKILL.md require a single credential: VOLCENGINE_API_KEY, which is proportionate to the declared purpose. However the registry metadata omitted required_env_vars, creating an inconsistency that may mislead users about what secrets are needed. No other unrelated credentials or broad environment access are requested.
Persistence & Privilege
The skill does not request always:true, does not claim persistent modification of other skills or system-wide settings, and uses no special privileges beyond normal file I/O and network access.
What to consider before installing
This skill's code implements the Seedream image-generation API and requires an API key (VOLCENGINE_API_KEY) even though the registry metadata omitted it. Before installing: 1) Confirm you are comfortable providing VOLCENGINE_API_KEY and that the key has appropriate, limited permissions; do not paste keys into code—use environment variables. 2) Understand the skill will upload any local image paths you pass (it base64-encodes and sends them to Volcengine) and will download whatever URLs the API returns into an output directory—avoid using sensitive images. 3) Verify the API endpoint domain (ark.cn-beijing.volces.com) is expected for your provider. 4) Install dependencies (requests, python-dotenv) in an isolated environment. 5) Because registry metadata omitted the required env var, consider reviewing the seedream_api.py source yourself or running it in a sandbox before granting the skill access to real API keys or sensitive files.

Like a lobster shell, security has layers — review code before you run it.

latestvk97bsq80jds7pjc551x5haeyqh82v2e0
301downloads
2stars
3versions
Updated 7h ago
v1.0.2
MIT-0

Seedream 图片生成 API 调用

此 skill 用于调用火山引擎 Seedream 图片生成 API,支持文生图功能。

使用方式: 用户只需提供图片描述,我直接调用 API 生成图片并返回本地文件。

支持的模型

代码中使用简短别名调用,完整 Model ID 如下:

名称别名Model ID说明
5.05.0doubao-seedream-5-0-260128默认使用,支持文生图/图生图/联网搜索
4.54.5doubao-seedream-4-5-251128支持文生图/图生图
4.04.0doubao-seedream-4-0-250828支持文生图/图生图
3.0-t2i3.0-t2idoubao-seedream-3-0-t2i-250415仅文生图
3.0-i2i3.0-i2idoubao-seededit-3-0-i2i-250415仅图生图

直接调用

用户说"生成一张xxx图片"时,直接运行:

python seedream_api.py "一只可爱的橘猫"

指定模型:

python seedream_api.py "一只可爱的橘猫" -m 4.5

图片会自动下载到 output 目录。

命令行参数

参数说明
prompt图片描述(位置参数)
-m, --model模型版本
-s, --size图片尺寸,默认 2048x2048
-o, --output-dir输出目录,默认 output
--steps推理步数 1-50,默认 50
--guidance引导系数 1-20,默认 7.5
--seed随机种子(可复现结果)
--negative负向提示词
-i, --image单张参考图片路径或URL(图生图)
--images多张参考图片路径或URL(多图生图/组图)
--group启用组图模式
--tools工具列表,如 web_search(启用联网搜索)

或导入使用:

from seedream_api import generate_image

result = generate_image("一只可爱的橘猫坐在窗台上")
for path in result["local_paths"]:
    print(path)  # 本地文件路径

API 端点

POST https://ark.cn-beijing.volces.com/api/v3/images/generations

鉴权方式

使用 API Key 进行鉴权,需要在请求头中添加:

Authorization: Bearer <your-api-key>
Content-Type: application/json

请求参数

必选参数

  • model (string): 模型名称

    • 使用代码中的别名:5.04.54.03.0-t2i
    • 或直接使用完整 Model ID
  • prompt (string): 文本提示词

    • 建议不超过 300 个汉字或 600 个英文单词
    • 字数过多可能导致模型忽略细节

可选参数

  • size (string): 生成图像的尺寸

    • 方式 1:指定分辨率(2K、4K)
    • 方式 2:指定宽高像素值(如 "2048x2048")
    • 总像素范围:[3686400, 16777216]
    • 宽高比范围:[1/16, 16]
    • 默认值:2048x2048

    推荐尺寸:

    • 2K 1:1: 2048x2048
    • 2K 16:9: 2848x1600
    • 2K 9:16: 1600x2848
    • 4K 1:1: 4096x4096
    • 4K 16:9: 5504x3040
    • 4K 9:16: 3040x5504
  • num_inference_steps (integer): 推理步数

    • 范围:1-50
    • 默认值:50
  • guidance_scale (number): 引导系数

    • 范围:1-20
    • 默认值:7.5
  • seed (integer): 随机种子

    • 用于生成可复现的图片
  • negative_prompt (string): 负向提示词

    • 描述不希望在图片中出现的内容
  • image (string/array): 参考图片信息

    • 支持文件路径、URL 或 Base64 编码
    • 单图:传入字符串
    • 多图:传入字符串数组(最多 14 张)
    • Base64 格式:data:image/<格式>;base64,<编码>,格式需小写
    • 支持格式:jpeg、png、webp、bmp、tiff、gif
    • 宽高比范围:[1/16, 16]
    • 宽高长度 > 14px
    • 大小:不超过 10MB
    • 总像素:不超过 6000x6000=36000000 px
  • sequential_image_generation (string): 组图模式

    • auto:启用组图模式,生成一组内容关联的图片
    • disabled:禁用组图模式,生成单张图片(默认)
    • 组图数量限制:最多 15 张(含参考图)
  • tools (array): 工具列表

    • 支持的工具类型:web_search(联网搜索)
    • 示例:[{"type": "web_search"}]
    • 用于增强提示词理解,模型可联网搜索相关信息
    • 仅支持 5.0 模型(doubao-seedream-5-0-260128)

使用场景

1. 文生图(生成单张图片)

{
  "model": "doubao-seedream-5-0-260128",
  "prompt": "一只可爱的橘猫坐在窗台上,阳光洒在它的毛发上",
  "size": "2048x2048"
}

2. 单图生图(图生图)

{
  "model": "doubao-seedream-5-0-260128",
  "prompt": "将图片转换成卡通风格",
  "image": "data:image/jpeg;base64,/9j/4AAQSkZJRg...",
  "size": "2048x2048"
}

3. 多图生图

{
  "model": "doubao-seedream-5-0-260128",
  "prompt": "结合这两张图片的风格",
  "image": [
    "data:image/jpeg;base64,/9j/4AAQSkZJRg...",
    "data:image/jpeg;base64,/9j/4AAQSkZJRg..."
  ],
  "size": "2048x2048"
}

4. 组图生成

{
  "model": "doubao-seedream-5-0-260128",
  "prompt": "一组不同姿势的橘猫",
  "sequential_image_generation": "auto",
  "size": "2048x2048"
}

5. 多图生组图

{
  "model": "doubao-seedream-5-0-260128",
  "prompt": "基于参考图片生成一组相关图片",
  "image": [
    "data:image/jpeg;base64,/9j/4AAQSkZJRg...",
    "data:image/jpeg;base64,/9j/4AAQSkZJRg..."
  ],
  "sequential_image_generation": "auto",
  "size": "2048x2048"
}

响应格式

成功响应:

{
  "created": 1710000000,
  "data": [
    {
      "url": "https://example.com/generated_image1.jpg"
    },
    {
      "url": "https://example.com/generated_image2.jpg"
    }
  ]
}

错误响应:

{
  "error": {
    "message": "错误信息",
    "type": "错误类型",
    "param": null,
    "code": null
  }
}

注意事项

  1. API Key 安全:不要在代码中硬编码 API Key,建议使用环境变量
  2. 图片 URL 可访问性:确保提供的图片 URL 可以被公开访问
  3. 提示词优化:简洁明确的提示词效果更好
  4. 尺寸限制:注意总像素和宽高比的双重限制
  5. 图片数量限制:最多支持 14 张参考图
  6. 格式要求:Base64 编码时注意格式正确,图片格式需小写
  7. 网络请求:代码会发起外部请求(API 请求和下载生成的图片),需确保网络畅通

Python 调用示例

import requests
import os

API_KEY = os.getenv("VOLCENGINE_API_KEY")
API_URL = "https://ark.cn-beijing.volces.com/api/v3/images/generations"

headers = {
    "Authorization": f"Bearer {API_KEY}",
    "Content-Type": "application/json"
}

data = {
    "model": "doubao-seedream-5-0-260128",
    "prompt": "一只可爱的橘猫坐在窗台上,阳光洒在它的毛发上",
    "size": "2048x2048"
}

response = requests.post(API_URL, headers=headers, json=data)
result = response.json()

if "data" in result:
    for image in result["data"]:
        print(f"生成的图片 URL: {image['url']}")
else:
    print(f"错误: {result.get('error', {}).get('message', '未知错误')}")

使用此 Skill

当用户需要以下功能时,使用此 skill:

  • 生成图片(文生图)
  • 图生图(使用参考图片)
  • 多图生图(多张参考图片)
  • 组图生成
  • 使用 Seedream 相关模型进行图片生成

Python 调用示例

文生图

from seedream_api import generate_image

result = generate_image("一只可爱的橘猫坐在窗台上")
for path in result["local_paths"]:
    print(path)  # 本地文件路径

图生图(使用文件路径)

from seedream_api import generate_image

result = generate_image(
    prompt="将图片转换成卡通风格",
    image="path/to/input.jpg"
)
for path in result["local_paths"]:
    print(path)

图生图(使用 URL)

from seedream_api import generate_image

result = generate_image(
    prompt="将图片转换成卡通风格",
    image="https://example.com/image.jpg"
)

多图生图

from seedream_api import generate_image

result = generate_image(
    prompt="结合这两张图片的风格",
    image=["path/to/image1.jpg", "path/to/image2.jpg"]
)

组图生成

from seedream_api import generate_image

result = generate_image(
    prompt="一组不同姿势的橘猫",
    sequential_image_generation="auto"
)
for path in result["local_paths"]:
    print(path)

使用联网搜索工具

from seedream_api import generate_image

result = generate_image(
    prompt="埃隆·马斯克最新的SpaceX火箭发射场景",
    tools=[{"type": "web_search"}]
)
for path in result["local_paths"]:
    print(path)

在调用前,确保:
1. 已获取火山引擎 API Key (环境变量: VOLCENGINE_API_KEY)
2. 已开通 Seedream 模型服务
3. 了解所需的使用场景和参数配置

Comments

Loading comments...