Description

使用可灵 Omni-Video API 生成或编辑视频。凡是涉及可灵视频的操作都应触发此 skill，包括但不限于：可灵生成视频、kling视频、文生视频(可灵)、图生视频(可灵)、视频编辑(可灵)、视频参考。当用户明确提到"可灵"或"kling"并需要视频生成/编辑时触发。配置：环境变量 HSAI_API_KEY。

README (SKILL.md)

可灵 Omni-Video 视频生成与编辑

Name: 可灵视频生成
Author: gayyzxyx

通过 scripts/generate_video.sh 完成视频全流程（提交任务 → 轮询状态 → 下载文件）。

使用方式

bash \x3Cskill-dir>/scripts/generate_video.sh \
  --prompt "视频描述" \
  [--model kling-video-o1] \
  [--output ./output.mp4] \
  [--aspect-ratio 16:9|9:16|1:1] \
  [--duration 5] \
  [--mode pro|std] \
  [--sound on|off] \
  [--image \x3Curl-or-base64>] \
  [--image-type first_frame|end_frame] \
  [--image2 \x3Curl-or-base64>] \
  [--image2-type first_frame|end_frame] \
  [--video \x3Curl>] \
  [--video-refer-type base|feature] \
  [--video-keep-sound yes|no] \
  [--poll-interval 10] \
  [--timeout 900]

参数说明

参数	必填	默认值	说明
`--prompt`	是	-	视频描述文本，最多 2500 字符。可用 `\x3C\x3C\x3Cimage_1>>>`、`\x3C\x3C\x3Cvideo_1>>>` 引用图片/视频
`--model`	否	`kling-v3-omni`	模型名称，可选 `kling-video-o1`
`--output`	否	`./kling_video_\x3Cts>.mp4`	输出文件路径
`--aspect-ratio`	条件	-	画面比例：`16:9`、`9:16`、`1:1`。纯文生视频时必填；有首帧图或视频编辑时不需要
`--duration`	否	`5`	时长（秒），可选 3-15。视频编辑模式下无效
`--mode`	否	`pro`	生成模式：`pro`（高品质）、`std`（标准，性价比高）
`--sound`	否	`off`	是否同时生成声音：`on`/`off`
`--image`	否	-	参考图片 URL（或 Base64）。可作为普通参考、首帧或尾帧
`--image-type`	否	-	图片帧类型：`first_frame` 或 `end_frame`，不设则为普通参考图
`--image2`	否	-	第二张参考图片 URL
`--image2-type`	否	-	第二张图片帧类型
`--video`	否	-	参考视频 URL（MP4/MOV，≤200MB）
`--video-refer-type`	否	`base`	视频类型：`base`（待编辑/指令变换）、`feature`（特征参考）
`--video-keep-sound`	否	`no`	是否保留视频原声：`yes`/`no`
`--poll-interval`	否	`10`	轮询间隔秒数
`--timeout`	否	`900`	最长等待秒数

常用场景

1. 文生视频

bash \x3Cskill-dir>/scripts/generate_video.sh \
  --prompt "一只猫在沙滩上奔跑" \
  --aspect-ratio "16:9" --duration 7 --mode pro

2. 图生视频（首帧）

bash \x3Cskill-dir>/scripts/generate_video.sh \
  --prompt "让画面中的人物向镜头挥手" \
  --image "https://example.com/photo.jpg" \
  --image-type first_frame --mode pro

3. 视频编辑（指令变换）

bash \x3Cskill-dir>/scripts/generate_video.sh \
  --prompt "给\x3C\x3C\x3Cvideo_1>>>中的人戴上墨镜" \
  --video "https://example.com/input.mp4" \
  --video-refer-type base --video-keep-sound yes

4. 视频参考（生成下一个镜头）

bash \x3Cskill-dir>/scripts/generate_video.sh \
  --prompt "基于\x3C\x3C\x3Cvideo_1>>>，生成下一个镜头" \
  --video "https://example.com/input.mp4" \
  --video-refer-type feature --aspect-ratio "16:9" --duration 5

注意事项

环境变量 HSAI_API_KEY 必须设置；依赖 curl 和 python3
视频生成耗时较长（通常 3-10 分钟），脚本会自动轮询等待
如果用户未指定 prompt，先向用户询问要生成什么内容的视频
如果用户提供的描述比较简短，可以帮助扩展为更详细的 prompt
有参考视频时，--sound 只能为 off
视频编辑模式（--video-refer-type base）下，输出视频时长与输入视频相同，--duration 和 --aspect-ratio 无效
视频生成完成后，告知用户文件路径，询问是否需要移动/复制

Usage Guidance

Before installing: 1) Treat this as an unverified third‑party integration — there is no homepage or author info. 2) The script requires HSAI_API_KEY (the registry metadata omits this) and will send that key to the configured API endpoint; verify you trust the endpoint. 3) The script also respects HSAI_BASE_URL (not documented in metadata) — do not set that to an untrusted host because it would receive your API key and prompt data. 4) If you proceed, restrict the API key's permissions and rotate it if misused; consider running the skill in an isolated environment or container and inspect network traffic to confirm it contacts the expected service host. 5) If you need higher assurance, ask the publisher for a homepage, docs, and clarity about the default API domain and why the registry metadata omits the required credential.

Capability Analysis

Type: OpenClaw Skill Name: kling-omni-video Version: 1.0.0 The skill bundle provides a legitimate interface for the Kling Omni-Video API, enabling video generation and editing. The shell script `scripts/generate_video.sh` correctly implements task submission, status polling, and file downloading using standard tools like `curl` and `python3` for JSON processing. The code follows security best practices (e.g., using `sys.argv` for Python arguments instead of string interpolation) and lacks any indicators of data exfiltration, backdoors, or malicious prompt injection.

Capability Assessment

⚠ Purpose & Capability

The skill's code and SKILL.md implement video generation against an Omni-Video API and require an API key (HSAI_API_KEY), which matches the described purpose. However the registry metadata lists no required environment variables or primary credential while both the SKILL.md and the script explicitly require HSAI_API_KEY — this metadata omission is an incoherence and reduces trust. The presence of a corp-like default API endpoint (https://api-aigw.corp.hongsong.club) and no homepage or publisher details is notable but not definitive proof of malicious intent.

ℹ Instruction Scope

SKILL.md instructs the agent to run scripts/generate_video.sh and the script confines itself to building a JSON payload, POSTing it to the API, polling task status, and downloading the returned video URL. The script does not read unrelated local files or other env vars. Caveat: the script prints prompt and metadata to stdout and will download a URL provided by the API (the returned VIDEO_URL), which could be controlled by the service — expected for this use case but worth noting.

✓ Install Mechanism

This is an instruction-only skill with a bundled script; there is no install step, no package downloads, and no archive extraction. Risk from install mechanism is low.

⚠ Credentials

The script requires HSAI_API_KEY (and SKILL.md mentions it) but the registry metadata failed to declare required env vars or a primary credential. Additionally the script honors an undocumented HSAI_BASE_URL environment variable that can redirect API calls to any endpoint (defaulting to a corp domain). That base-URL override can be abused to send your API key and prompt data to an attacker-controlled server if set; the presence of this override without explicit documentation is a proportionality and transparency concern.

✓ Persistence & Privilege

The skill is not always: true, does not request system-wide configuration changes, and does not modify other skills. Autonomous invocation is allowed (platform default) but not combined with other high-risk privileges in this package.

Version History

v1.0.0

Initial release: 可灵 Omni-Video API 视频生成与编辑

Metadata

Slug kling-omni-video

Version 1.0.0

License MIT-0

All-time Installs 0

Active Installs 0

Total Versions 1

Frequently Asked Questions

What is 可灵视频生成?

使用可灵 Omni-Video API 生成或编辑视频。凡是涉及可灵视频的操作都应触发此 skill，包括但不限于：可灵生成视频、kling视频、文生视频(可灵)、图生视频(可灵)、视频编辑(可灵)、视频参考。当用户明确提到"可灵"或"kling"并需要视频生成/编辑时触发。配置：环境变量 HSAI_API_KEY。 It is an AI Agent Skill for Claude Code / OpenClaw, with 77 downloads so far.

How do I install 可灵视频生成?

Run "/install kling-omni-video" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is 可灵视频生成 free?

Yes, 可灵视频生成 is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does 可灵视频生成 support?

可灵视频生成 is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created 可灵视频生成?

It is built and maintained by gayyzxyx (@gayyzxyx); the current version is v1.0.0.

More Skills

可灵视频生成