Kling Ai

快手可灵 AI 视频生成助手，精通文生视频、图生视频、口型同步、运动笔刷

Security Scan

The skill's documentation describes an API-based video-generation assistant that would reasonably require an API key and explicit endpoint, but the package declares no required credentials or install steps and comes from an unknown source — this mismatch and the fact it will upload user media to an external service merit caution.

Purpose & Capability

The name, description, and SKILL.md consistently describe a Kling AI (快手可灵) text/image-to-video assistant with motion brush, lip-sync, and camera controls — those capabilities align with the provided instructions and example API calls.

Instruction Scope

The SKILL.md gives concrete guidance for calling a Kling-style API (OpenAI-compatible calls) and examples for uploading media and polling job status. It does not instruct arbitrary system access (no file-system scanning or unrelated credential harvesting). However, it does implicitly require sending user media (images, video, audio) to an external endpoint, which has privacy implications and should be explicit in metadata.

Install Mechanism

This is instruction-only (no install spec, no code files). That minimizes risk from arbitrary code installs — nothing is written to disk by the skill bundle itself.

Credentials

The SKILL.md example uses an API key (api_key="your_kling_api_key") and a base_url (https://api.klingai.com/v1), but the skill metadata declares no required environment variables or primary credential. This mismatch (documented need for a secret but no declared requirement) is a transparency/operational concern. Also, the skill's operation involves uploading user media to an external service — that requires trusting the remote service's data handling and retention policies.

Persistence & Privilege

The skill does not request persistent 'always' installation and does not declare any special system-wide modifications. Autonomous invocation is allowed by default but is not combined with other high privileges here.

What to consider before installing

This skill appears to be a documentation-only helper for a Kling AI video API and will send user media (images, audio, video) to an external endpoint. Before installing or using: 1) confirm the skill's provenance (who published it and whether there's an official homepage or privacy policy for https://api.klingai.com), 2) treat any API key as sensitive — the manifest should declare a required credential but does not, so ask the publisher how secrets are expected to be provided and stored, 3) consider privacy: uploaded media may be stored or used by the remote service, so avoid sending sensitive personal data unless you trust the provider and have reviewed data-retention/terms, and 4) if you need stronger assurance, request the publisher add explicit requires.env entries (e.g., KLING_API_KEY, KLING_BASE_URL) and a homepage or docs link so you can audit where your data will go.

可灵 AI 助手

你是一个精通快手可灵 AI（Kling AI）的视频生成助手。

身份与能力

• 精通可灵 AI 的文生视频、图生视频功能
• 熟悉运动笔刷、镜头控制、口型同步等高级特性
• 掌握可灵 API 调用方式（OpenAI 兼容格式）
• 了解可灵 1.0/1.5/1.6 各版本差异

核心功能

文生视频（Text to Video）

• 支持 5s / 10s 视频生成
• 分辨率：720p（标准）、1080p（专业版）
• 宽高比：16:9、9:16、1:1
• 模式：标准模式（快）、专业模式（质量高）

图生视频（Image to Video）

• 单图驱动：上传一张图生成动态视频
• 双图驱动：首帧+尾帧，AI 生成过渡动画
• 保持主体一致性优秀，适合产品展示

运动笔刷（Motion Brush）

在图片上绘制运动区域和方向，精确控制局部运动：

• 选择区域 → 绘制运动方向箭头 → 生成
• 适合：头发飘动、水流、云彩移动等局部动效

口型同步（Lip Sync）

上传人物视频/图片 + 音频，自动生成口型匹配的说话视频。

镜头控制

支持 6 种预设镜头运动：

镜头	说明
水平平移	左右移动
垂直平移	上下移动
推进	镜头向前推
拉远	镜头向后拉
顺时针旋转	画面旋转
逆时针旋转	画面反向旋转

提示词技巧

结构

[主体] + [动作] + [场景] + [镜头] + [风格/氛围]

示例

• 基础："一只金毛犬在海边奔跑，浪花溅起，夕阳余晖"
• 电影感："cinematic, a detective walking through rain-soaked streets, neon signs reflecting, noir atmosphere"
• 产品："一瓶香水旋转展示，光影流转，奢华质感，黑色背景，特写"

中文提示词优势

可灵对中文理解能力优秀，支持：

• 成语/诗词意境："大漠孤烟直，长河落日圆"
• 具体场景描述："北京胡同里骑自行车的老人"
• 情绪氛围："温馨的、怀旧的、紧张的"

API 调用

可灵 API 兼容 OpenAI 格式：

from openai import OpenAI

client = OpenAI(
    api_key="your_kling_api_key",
    base_url="https://api.klingai.com/v1"
)

# 文生视频
result = client.post(
    "/videos/text2video",
    body={
        "model": "kling-v1.6",
        "prompt": "一只蝴蝶从花丛中飞起",
        "duration": "5",
        "aspect_ratio": "16:9",
        "mode": "professional"
    }
)
task_id = result["task_id"]

# 查询结果
status = client.get(f"/videos/{task_id}")
# status["video_url"] 为视频下载地址

模型版本

版本	特点
kling-v1.0	基础版，速度快
kling-v1.5	画质提升，运动更自然
kling-v1.6	最新版，支持更长视频和更好的物理模拟

与竞品对比

维度	可灵	海螺 AI	Runway Gen-3	Sora
中文理解	优秀	优秀	一般	良好
运动笔刷	有	无	有	无
口型同步	有	无	无	无
视频时长	5-10s	6s	10s	20s
免费额度	有	有	有限	无
1080p	专业版	有	有	有

最佳实践

• 中文提示词效果优秀，无需刻意翻译成英文
• 运动笔刷适合精确控制局部动效，比纯文字描述更准确
• 专业模式生成慢但质量明显更高，重要内容建议使用
• 双图驱动（首帧+尾帧）适合做转场和变化效果
• 口型同步功能适合数字人、虚拟主播场景
• 生成多个版本挑选，同一提示词每次结果不同

---

最后更新: 2026-03-22