Kling Image Generate

可灵AI图像生成API工具。支持文生图、图生图、多图参考生成、图像Omni、扩图等功能。使用环境变量KLING_ACCESS_KEY和KLING_SECRET_KEY进行鉴权。当用户需要生成AI图像、图片编辑、图像扩展等任务时使用此技能。

Security Scan

The skill's code and runtime instructions match an image-generation client that legitimately needs KLING_ACCESS_KEY/KLING_SECRET_KEY, but the registry metadata omits those required secrets and there are a few minor scope/operational concerns you should review before installing.

Purpose & Capability

The scripts and SKILL.md implement a client for the Kling AI image API (text->image, image->image, omni, expansion). The requested credentials (KLING_ACCESS_KEY, KLING_SECRET_KEY) and dependencies (requests, JWT libraries) are coherent with that purpose. However, registry metadata lists no required env vars/primary credential while the SKILL.md and all scripts clearly require the two KLING_* environment variables — this metadata mismatch is an inconsistency to be aware of.

Instruction Scope

SKILL.md instructs users to set the two KLING_* env vars and run included Python scripts; the scripts only call the documented api-beijing.klingai.com endpoints and do not attempt to read unrelated system files or other credentials. One operational note: the scripts accept an optional callback_url parameter (which will be invoked by the remote service) — if you supply a third-party callback URL you could direct generated results to an external endpoint, so treat callback_url as a potential exfil vector if misused.

Install Mechanism

No install specification is included (instruction- + script bundle). The included requirements.txt (requests, PyJWT, cryptography) matches the code's usage. There are no external downloads or obscure install URLs in the package.

Credentials

The skill requires KLING_ACCESS_KEY and KLING_SECRET_KEY for JWT authentication — this is appropriate and necessary for the stated API calls. The concern is that the package/registry metadata omitted these required env variables (declared as 'none'), which is an internal inconsistency and could mislead less-technical users into installing without supplying keys. No other unrelated secrets are requested.

Persistence & Privilege

The skill does not request persistent 'always' inclusion and does not modify other skills or system-wide configuration. It runs as user-invoked scripts and has normal, limited privileges.

What to consider before installing

Before installing: (1) Note that the package requires KLING_ACCESS_KEY and KLING_SECRET_KEY (the registry metadata incorrectly omitted them) — only provide keys you trust and consider using a dedicated/limited account. (2) Confirm you trust the API host api-beijing.klingai.com and are comfortable the service will receive JWT-authenticated requests. (3) Avoid supplying an untrusted callback_url (it can cause results to be posted to an external endpoint). (4) Install dependencies in an isolated environment (virtualenv) and review the scripts yourself; if you later suspect key leakage, rotate the keys. If you want higher assurance, ask the publisher for a homepage or source repo and a corrected metadata entry that lists the required env vars.

可灵图像生成

🇨🇳 中文 | 🇬🇧 English

可灵AI图像生成服务，提供文生图、图生图、扩图等多种图像生成和编辑能力。

🔒 安全说明: 本技能需要调用可灵AI官方 API (api-beijing.klingai.com)，并使用用户提供的 API Key 进行 JWT 鉴权。所有凭证仅存储在本地环境变量中，不会上传或共享。

快速开始

1. 配置环境变量

export KLING_ACCESS_KEY="your_access_key"
export KLING_SECRET_KEY="your_secret_key"

2. 运行脚本

# 文生图（推荐带进度版本）
python3 scripts/generate_image_with_progress.py \
  --prompt "一只可爱的猫咪，皮克斯风格" \
  --model kling-v3 \
  --n 2 \
  --aspect_ratio 1:1 \
  --wait

# 图生图
python3 scripts/generate_image_with_progress.py \
  --prompt "保持原图风格，添加花朵装饰" \
  --image "https://example.com/image.png" \
  --image_reference subject \
  --wait

脚本清单

脚本	用途	推荐度
generate_image_with_progress.py	图像生成（带进度显示）	⭐ 推荐
generate_image.py	图像生成（基础版）	可选
generate_omni_image.py	Omni多图生成	按需
expand_image.py	图像扩图	按需
query_task.py	查询任务状态	工具
list_tasks.py	获取任务列表	工具

核心功能使用

文生图

python3 scripts/generate_image_with_progress.py \
  --prompt "描述你想要的图像内容" \
  --model kling-v3 \
  --n 1 \
  --aspect_ratio 1:1 \
  --resolution 2k \
  --wait

常用参数：

• --model: 模型（kling-v3, kling-v2-1, kling-v1-5 等）
• --n: 生成数量 1-9
• --aspect_ratio: 1:1, 16:9, 9:16, 4:3 等
• --resolution: 1k, 2k
• --wait: 等待任务完成并显示进度

图生图

python3 scripts/generate_image_with_progress.py \
  --prompt "基于参考图的描述" \
  --image "https://example.com/ref.jpg" \
  --image_reference subject \
  --image_fidelity 0.7 \
  --wait

参考类型：

• subject: 主体参考（保持人物/物体主体）
• face: 面部参考（保持面部特征）

Omni多图生成

支持多图参考、风格迁移等高级功能：

python3 scripts/generate_omni_image.py \
  --prompt "将<<<image_1>>>的风格应用到<<<image_2>>>上" \
  --images "url1,url2" \
  --model kling-v3-omni \
  --resolution 2k \
  --wait

扩图

智能扩展图像边界：

# 手动指定扩展比例
python3 scripts/expand_image.py \
  --image "https://example.com/img.jpg" \
  --up 0.5 --down 0.5 --left 0.5 --right 0.5 \
  --prompt "扩展区域的描述" \
  --wait

# 自动计算比例
python3 scripts/expand_image.py \
  --image "https://example.com/img.jpg" \
  --auto_ratio \
  --width 1024 --height 1024 \
  --aspect_ratio 16:9 \
  --wait

任务管理

查询任务状态

python3 scripts/query_task.py --task_id "xxx"

获取任务列表

python3 scripts/list_tasks.py --api_type generation --page 1

详细参考

• 完整 API 参数: 参见 references/api-reference.md
• English Version: SKILL.en.md